Warning: Permanently added '2620:52:3:1:dead:beef:cafe:c116' (ED25519) to the list of known hosts.

You can reproduce this build on your computer by running:

  sudo dnf install copr-rpmbuild
  /usr/bin/copr-rpmbuild --verbose --drop-resultdir --task-url https://copr.fedorainfracloud.org/backend/get-build-task/8319797-fedora-rawhide-x86_64 --chroot fedora-rawhide-x86_64


Version: 1.2
PID: 8875
Logging PID: 8876
Task:
{'allow_user_ssh': False,
 'appstream': False,
 'background': True,
 'build_id': 8319797,
 'buildroot_pkgs': [],
 'chroot': 'fedora-rawhide-x86_64',
 'enable_net': False,
 'fedora_review': False,
 'git_hash': '0d4b6a9c8268aaf632945aaeee9940f5896f97ad',
 'git_repo': 'https://copr-dist-git.fedorainfracloud.org/git/@rocm-packagers-sig/RH:pr:12/llama-cpp',
 'isolation': 'default',
 'memory_reqs': 2048,
 'package_name': 'llama-cpp',
 'package_version': 'b4094-6',
 'project_dirname': 'RH:pr:12',
 'project_name': 'RH',
 'project_owner': '@rocm-packagers-sig',
 'repo_priority': None,
 'repos': [{'baseurl': 'https://download.copr.fedorainfracloud.org/results/@rocm-packagers-sig/RH/fedora-rawhide-x86_64/',
            'id': 'copr_base',
            'name': 'Copr repository',
            'priority': None}],
 'sandbox': '@rocm-packagers-sig/RH--https://src.fedoraproject.org/user/rishi',
 'source_json': {},
 'source_type': None,
 'ssh_public_keys': None,
 'storage': 0,
 'submitter': 'https://src.fedoraproject.org/user/rishi',
 'tags': [],
 'task_id': '8319797-fedora-rawhide-x86_64',
 'timeout': 18000,
 'uses_devel_repo': False,
 'with_opts': [],
 'without_opts': []}

Running: git clone https://copr-dist-git.fedorainfracloud.org/git/@rocm-packagers-sig/RH:pr:12/llama-cpp /var/lib/copr-rpmbuild/workspace/workdir-a9lx8hdc/llama-cpp --depth 500 --no-single-branch --recursive

cmd: ['git', 'clone', 'https://copr-dist-git.fedorainfracloud.org/git/@rocm-packagers-sig/RH:pr:12/llama-cpp', '/var/lib/copr-rpmbuild/workspace/workdir-a9lx8hdc/llama-cpp', '--depth', '500', '--no-single-branch', '--recursive']
cwd: .
rc: 0
stdout: 
stderr: Cloning into '/var/lib/copr-rpmbuild/workspace/workdir-a9lx8hdc/llama-cpp'...

Running: git checkout 0d4b6a9c8268aaf632945aaeee9940f5896f97ad --

cmd: ['git', 'checkout', '0d4b6a9c8268aaf632945aaeee9940f5896f97ad', '--']
cwd: /var/lib/copr-rpmbuild/workspace/workdir-a9lx8hdc/llama-cpp
rc: 0
stdout: 
stderr: Note: switching to '0d4b6a9c8268aaf632945aaeee9940f5896f97ad'.

You are in 'detached HEAD' state. You can look around, make experimental
changes and commit them, and you can discard any commits you make in this
state without impacting any branches by switching back to a branch.

If you want to create a new branch to retain commits you create, you may
do so (now or later) by using -c with the switch command. Example:

  git switch -c <new-branch-name>

Or undo this operation with:

  git switch -

Turn off this advice by setting config variable advice.detachedHead to false

HEAD is now at 0d4b6a9 automatic import of llama-cpp

Running: dist-git-client sources

cmd: ['dist-git-client', 'sources']
cwd: /var/lib/copr-rpmbuild/workspace/workdir-a9lx8hdc/llama-cpp
rc: 0
stdout: 
stderr: INFO: Reading stdout from command: git rev-parse --abbrev-ref HEAD
INFO: Reading stdout from command: git rev-parse HEAD
INFO: Reading sources specification file: sources
INFO: Downloading llama.cpp-b4094.tar.gz
INFO: Reading stdout from command: curl --help all
INFO: Calling: curl -H Pragma: -o llama.cpp-b4094.tar.gz --location --connect-timeout 60 --retry 3 --retry-delay 10 --remote-time --show-error --fail --retry-all-errors https://copr-dist-git.fedorainfracloud.org/repo/pkgs/@rocm-packagers-sig/RH:pr:12/llama-cpp/llama.cpp-b4094.tar.gz/md5/b15abcdba81002328c19aa1c85077c33/llama.cpp-b4094.tar.gz
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100 18.6M  100 18.6M    0     0  75.5M      0 --:--:-- --:--:-- --:--:-- 75.5M
INFO: Reading stdout from command: md5sum llama.cpp-b4094.tar.gz

/usr/bin/tail: /var/lib/copr-rpmbuild/main.log: file truncated
Running (timeout=18000): unbuffer mock --spec /var/lib/copr-rpmbuild/workspace/workdir-a9lx8hdc/llama-cpp/llama-cpp.spec --sources /var/lib/copr-rpmbuild/workspace/workdir-a9lx8hdc/llama-cpp --resultdir /var/lib/copr-rpmbuild/results --uniqueext 1732737731.306783 -r /var/lib/copr-rpmbuild/results/configs/child.cfg
INFO: mock.py version 5.9 starting (python version = 3.13.0, NVR = mock-5.9-1.fc41), args: /usr/libexec/mock/mock --spec /var/lib/copr-rpmbuild/workspace/workdir-a9lx8hdc/llama-cpp/llama-cpp.spec --sources /var/lib/copr-rpmbuild/workspace/workdir-a9lx8hdc/llama-cpp --resultdir /var/lib/copr-rpmbuild/results --uniqueext 1732737731.306783 -r /var/lib/copr-rpmbuild/results/configs/child.cfg
Start(bootstrap): init plugins
INFO: tmpfs initialized
INFO: selinux enabled
INFO: chroot_scan: initialized
INFO: compress_logs: initialized
Finish(bootstrap): init plugins
Start: init plugins
INFO: tmpfs initialized
INFO: selinux enabled
INFO: chroot_scan: initialized
INFO: compress_logs: initialized
Finish: init plugins
INFO: Signal handler active
Start: run
INFO: Start(/var/lib/copr-rpmbuild/workspace/workdir-a9lx8hdc/llama-cpp/llama-cpp.spec)  Config(fedora-rawhide-x86_64)
Start: clean chroot
Finish: clean chroot
Mock Version: 5.9
INFO: Mock Version: 5.9
Start(bootstrap): chroot init
INFO: mounting tmpfs at /var/lib/mock/fedora-rawhide-x86_64-bootstrap-1732737731.306783/root.
INFO: calling preinit hooks
INFO: enabled root cache
INFO: enabled package manager cache
Start(bootstrap): cleaning package manager metadata
Finish(bootstrap): cleaning package manager metadata
INFO: Guessed host environment type: unknown
INFO: Using bootstrap image: registry.fedoraproject.org/fedora:rawhide
INFO: Pulling image: registry.fedoraproject.org/fedora:rawhide
INFO: Copy content of container registry.fedoraproject.org/fedora:rawhide to /var/lib/mock/fedora-rawhide-x86_64-bootstrap-1732737731.306783/root
INFO: Checking that registry.fedoraproject.org/fedora:rawhide image matches host's architecture
INFO: mounting registry.fedoraproject.org/fedora:rawhide with podman image mount
INFO: image registry.fedoraproject.org/fedora:rawhide as /var/lib/containers/storage/overlay/ddfc0dda31213fca22a3aefbee5638d5ad3ed32301368e52b3d7b89b528dfee0/merged
INFO: umounting image registry.fedoraproject.org/fedora:rawhide (/var/lib/containers/storage/overlay/ddfc0dda31213fca22a3aefbee5638d5ad3ed32301368e52b3d7b89b528dfee0/merged) with podman image umount
INFO: Package manager dnf5 detected and used (fallback)
INFO: Not updating bootstrap chroot, bootstrap_image_ready=True
Start(bootstrap): creating root cache
Finish(bootstrap): creating root cache
Finish(bootstrap): chroot init
Start: chroot init
INFO: mounting tmpfs at /var/lib/mock/fedora-rawhide-x86_64-1732737731.306783/root.
INFO: calling preinit hooks
INFO: enabled root cache
INFO: enabled package manager cache
Start: cleaning package manager metadata
Finish: cleaning package manager metadata
INFO: enabled HW Info plugin
INFO: Package manager dnf5 detected and used (direct choice)
INFO: Buildroot is handled by package management downloaded with a bootstrap image:
  rpm-4.20.0-1.fc42.x86_64
  rpm-sequoia-1.7.0-2.fc41.x86_64
  dnf5-5.2.7.0-1.fc42.x86_64
  dnf5-plugins-5.2.7.0-1.fc42.x86_64
Start: installing minimal buildroot with dnf5
Updating and loading repositories:
 fedora                                 100% |  20.0 MiB/s |  22.8 MiB |  00m01s
 Copr repository                        100% | 732.4 KiB/s |  60.1 KiB |  00m00s
Repositories loaded.
Package                            Arch   Version                    Repository      Size
Installing group/module packages:
 bash                              x86_64 5.2.37-1.fc42              fedora       8.2 MiB
 bzip2                             x86_64 1.0.8-19.fc41              fedora      95.7 KiB
 coreutils                         x86_64 9.5-11.fc42                fedora       5.4 MiB
 cpio                              x86_64 2.15-2.fc41                fedora       1.1 MiB
 diffutils                         x86_64 3.10-8.fc41                fedora       1.6 MiB
 fedora-release-common             noarch 42-0.8                     fedora      19.7 KiB
 findutils                         x86_64 1:4.10.0-4.fc41            fedora       1.8 MiB
 gawk                              x86_64 5.3.0-4.fc41               fedora       1.7 MiB
 glibc-minimal-langpack            x86_64 2.40.9000-20.fc42          fedora       0.0   B
 grep                              x86_64 3.11-9.fc41                fedora       1.0 MiB
 gzip                              x86_64 1.13-2.fc41                fedora     389.0 KiB
 info                              x86_64 7.1.1-2.fc42               fedora     361.8 KiB
 patch                             x86_64 2.7.6-25.fc41              fedora     266.7 KiB
 redhat-rpm-config                 noarch 296-1.fc42                 fedora     186.6 KiB
 rpm-build                         x86_64 4.20.0-1.fc42              fedora     194.3 KiB
 sed                               x86_64 4.9-3.fc41                 fedora     861.5 KiB
 shadow-utils                      x86_64 2:4.16.0-7.fc42            fedora       4.0 MiB
 tar                               x86_64 2:1.35-4.fc41              fedora       2.9 MiB
 unzip                             x86_64 6.0-65.fc42                fedora     398.2 KiB
 util-linux                        x86_64 2.40.2-8.fc42              fedora       3.7 MiB
 which                             x86_64 2.21-42.fc41               fedora      80.2 KiB
 xz                                x86_64 1:5.6.3-2.fc42             fedora       1.2 MiB
Installing dependencies:
 add-determinism                   x86_64 0.4.3-1.fc42               fedora       2.4 MiB
 alternatives                      x86_64 1.30-1.fc41                fedora      66.3 KiB
 ansible-srpm-macros               noarch 1-16.fc41                  fedora      35.7 KiB
 audit-libs                        x86_64 4.0.2-1.fc41               fedora     331.3 KiB
 authselect                        x86_64 1.5.0-8.fc42               fedora     157.5 KiB
 authselect-libs                   x86_64 1.5.0-8.fc42               fedora     822.2 KiB
 basesystem                        noarch 11-21.fc41                 fedora       0.0   B
 binutils                          x86_64 2.43.50-8.fc42             fedora      25.4 MiB
 build-reproducibility-srpm-macros noarch 0.4.3-1.fc42               fedora     735.0   B
 bzip2-libs                        x86_64 1.0.8-19.fc41              fedora      80.7 KiB
 ca-certificates                   noarch 2024.2.69_v8.0.401-2.fc42  fedora       2.6 MiB
 coreutils-common                  x86_64 9.5-11.fc42                fedora      11.2 MiB
 cracklib                          x86_64 2.9.11-6.fc41              fedora     238.9 KiB
 crypto-policies                   noarch 20241106-1.git35892de.fc42 fedora     136.9 KiB
 curl                              x86_64 8.10.1-2.fc42              fedora     453.3 KiB
 cyrus-sasl-lib                    x86_64 2.1.28-27.fc41             fedora       2.3 MiB
 debugedit                         x86_64 5.1-1.fc42                 fedora     192.7 KiB
 dwz                               x86_64 0.15-8.fc42                fedora     299.2 KiB
 ed                                x86_64 1.20.2-2.fc41              fedora     146.9 KiB
 efi-srpm-macros                   noarch 5-13.fc42                  fedora      40.2 KiB
 elfutils                          x86_64 0.192-6.fc42               fedora       2.6 MiB
 elfutils-debuginfod-client        x86_64 0.192-6.fc42               fedora      80.4 KiB
 elfutils-default-yama-scope       noarch 0.192-6.fc42               fedora       1.8 KiB
 elfutils-libelf                   x86_64 0.192-6.fc42               fedora       1.2 MiB
 elfutils-libs                     x86_64 0.192-6.fc42               fedora     672.8 KiB
 fedora-gpg-keys                   noarch 42-0.3                     fedora     126.4 KiB
 fedora-release                    noarch 42-0.8                     fedora       0.0   B
 fedora-release-identity-basic     noarch 42-0.8                     fedora     719.0   B
 fedora-repos                      noarch 42-0.3                     fedora       4.9 KiB
 fedora-repos-rawhide              noarch 42-0.3                     fedora       2.2 KiB
 file                              x86_64 5.45-8.fc42                fedora     103.7 KiB
 file-libs                         x86_64 5.45-8.fc42                fedora       9.9 MiB
 filesystem                        x86_64 3.18-29.fc42               fedora     106.0   B
 fonts-srpm-macros                 noarch 1:2.0.5-17.fc41            fedora      55.8 KiB
 forge-srpm-macros                 noarch 0.4.0-1.fc42               fedora      38.9 KiB
 fpc-srpm-macros                   noarch 1.3-13.fc41                fedora     144.0   B
 gdb-minimal                       x86_64 15.2-3.fc42                fedora      13.0 MiB
 gdbm                              x86_64 1:1.23-7.fc41              fedora     460.9 KiB
 gdbm-libs                         x86_64 1:1.23-7.fc41              fedora     121.9 KiB
 ghc-srpm-macros                   noarch 1.9.2-1.fc42               fedora     779.0   B
 glibc                             x86_64 2.40.9000-20.fc42          fedora       6.7 MiB
 glibc-common                      x86_64 2.40.9000-20.fc42          fedora       1.1 MiB
 glibc-gconv-extra                 x86_64 2.40.9000-20.fc42          fedora       8.0 MiB
 gmp                               x86_64 1:6.3.0-2.fc41             fedora     811.4 KiB
 gnat-srpm-macros                  noarch 6-6.fc41                   fedora       1.0 KiB
 go-srpm-macros                    noarch 3.6.0-3.fc41               fedora      60.8 KiB
 jansson                           x86_64 2.14-1.fc42                fedora      93.1 KiB
 json-c                            x86_64 0.18-1.fc42                fedora      83.3 KiB
 kernel-srpm-macros                noarch 1.0-24.fc41                fedora       1.9 KiB
 keyutils-libs                     x86_64 1.6.3-4.fc41               fedora      54.4 KiB
 krb5-libs                         x86_64 1.21.3-3.fc42              fedora       2.3 MiB
 libacl                            x86_64 2.3.2-2.fc41               fedora      40.0 KiB
 libarchive                        x86_64 3.7.7-1.fc42               fedora     932.3 KiB
 libattr                           x86_64 2.5.2-4.fc41               fedora      28.5 KiB
 libblkid                          x86_64 2.40.2-8.fc42              fedora     262.5 KiB
 libbrotli                         x86_64 1.1.0-5.fc41               fedora     837.6 KiB
 libcap                            x86_64 2.71-1.fc42                fedora     210.8 KiB
 libcap-ng                         x86_64 0.8.5-3.fc41               fedora      69.2 KiB
 libcom_err                        x86_64 1.47.1-6.fc42              fedora      67.2 KiB
 libcurl                           x86_64 8.10.1-2.fc42              fedora     838.4 KiB
 libeconf                          x86_64 0.7.4-3.fc42               fedora      65.7 KiB
 libevent                          x86_64 2.1.12-14.fc41             fedora     895.7 KiB
 libfdisk                          x86_64 2.40.2-8.fc42              fedora     362.9 KiB
 libffi                            x86_64 3.4.6-3.fc42               fedora      86.4 KiB
 libgcc                            x86_64 14.2.1-6.fc42              fedora     270.6 KiB
 libgomp                           x86_64 14.2.1-6.fc42              fedora     519.8 KiB
 libidn2                           x86_64 2.3.7-2.fc41               fedora     329.1 KiB
 libmount                          x86_64 2.40.2-8.fc42              fedora     355.8 KiB
 libnghttp2                        x86_64 1.64.0-1.fc42              fedora     174.5 KiB
 libnsl2                           x86_64 2.0.1-2.fc41               fedora      57.9 KiB
 libpkgconf                        x86_64 2.3.0-1.fc42               fedora      78.2 KiB
 libpsl                            x86_64 0.21.5-4.fc41              fedora      80.5 KiB
 libpwquality                      x86_64 1.4.5-11.fc41              fedora     417.8 KiB
 libselinux                        x86_64 3.7-6.fc42                 fedora     181.1 KiB
 libsemanage                       x86_64 3.7-3.fc42                 fedora     298.0 KiB
 libsepol                          x86_64 3.7-3.fc42                 fedora     818.0 KiB
 libsmartcols                      x86_64 2.40.2-8.fc42              fedora     180.4 KiB
 libssh                            x86_64 0.11.1-1.fc42              fedora     569.6 KiB
 libssh-config                     noarch 0.11.1-1.fc42              fedora     277.0   B
 libstdc++                         x86_64 14.2.1-6.fc42              fedora       2.8 MiB
 libtasn1                          x86_64 4.19.0-9.fc41              fedora     175.7 KiB
 libtirpc                          x86_64 1.3.6-1.fc42               fedora     205.5 KiB
 libtool-ltdl                      x86_64 2.4.7-12.fc41              fedora      66.2 KiB
 libunistring                      x86_64 1.1-8.fc41                 fedora       1.7 MiB
 libuuid                           x86_64 2.40.2-8.fc42              fedora      41.4 KiB
 libverto                          x86_64 0.3.2-9.fc41               fedora      29.5 KiB
 libxcrypt                         x86_64 4.4.36-10.fc42             fedora     267.7 KiB
 libxml2                           x86_64 2.12.8-2.fc41              fedora       1.7 MiB
 libzstd                           x86_64 1.5.6-2.fc41               fedora     795.9 KiB
 lua-libs                          x86_64 5.4.7-1.fc42               fedora     285.0 KiB
 lua-srpm-macros                   noarch 1-14.fc41                  fedora       1.3 KiB
 lz4-libs                          x86_64 1.10.0-1.fc41              fedora     145.5 KiB
 mpfr                              x86_64 4.2.1-5.fc41               fedora     832.1 KiB
 ncurses-base                      noarch 6.5-2.20240629.fc41        fedora     326.3 KiB
 ncurses-libs                      x86_64 6.5-2.20240629.fc41        fedora     975.2 KiB
 ocaml-srpm-macros                 noarch 10-3.fc41                  fedora       1.9 KiB
 openblas-srpm-macros              noarch 2-18.fc41                  fedora     112.0   B
 openldap                          x86_64 2.6.8-5.fc41               fedora     644.2 KiB
 openssl-libs                      x86_64 1:3.2.2-8.fc42             fedora       7.8 MiB
 p11-kit                           x86_64 0.25.5-4.fc42              fedora       2.2 MiB
 p11-kit-trust                     x86_64 0.25.5-4.fc42              fedora     403.8 KiB
 package-notes-srpm-macros         noarch 0.5-12.fc41                fedora       1.6 KiB
 pam                               x86_64 1.7.0-3.fc42               fedora       1.8 MiB
 pam-libs                          x86_64 1.7.0-3.fc42               fedora     139.4 KiB
 pcre2                             x86_64 10.44-1.fc41.1             fedora     653.5 KiB
 pcre2-syntax                      noarch 10.44-1.fc41.1             fedora     251.6 KiB
 perl-srpm-macros                  noarch 1-56.fc41                  fedora     861.0   B
 pkgconf                           x86_64 2.3.0-1.fc42               fedora      88.6 KiB
 pkgconf-m4                        noarch 2.3.0-1.fc42               fedora      14.4 KiB
 pkgconf-pkg-config                x86_64 2.3.0-1.fc42               fedora     989.0   B
 popt                              x86_64 1.19-7.fc41                fedora     136.9 KiB
 publicsuffix-list-dafsa           noarch 20240107-4.fc41            fedora      67.5 KiB
 pyproject-srpm-macros             noarch 1.16.2-1.fc42              fedora       1.9 KiB
 python-srpm-macros                noarch 3.13-3.fc41                fedora      51.0 KiB
 qt5-srpm-macros                   noarch 5.15.15-1.fc42             fedora     500.0   B
 qt6-srpm-macros                   noarch 6.8.0-1.fc42               fedora     456.0   B
 readline                          x86_64 8.2-11.fc42                fedora     493.1 KiB
 rpm                               x86_64 4.20.0-1.fc42              fedora       3.1 MiB
 rpm-build-libs                    x86_64 4.20.0-1.fc42              fedora     206.7 KiB
 rpm-libs                          x86_64 4.20.0-1.fc42              fedora     726.1 KiB
 rpm-sequoia                       x86_64 1.7.0-2.fc41               fedora       2.4 MiB
 rust-srpm-macros                  noarch 26.3-3.fc42                fedora       4.8 KiB
 setup                             noarch 2.15.0-5.fc41              fedora     720.7 KiB
 sqlite-libs                       x86_64 3.47.0-1.fc42              fedora       1.5 MiB
 systemd-libs                      x86_64 257~rc2-2.fc42             fedora       2.3 MiB
 util-linux-core                   x86_64 2.40.2-8.fc42              fedora       1.5 MiB
 xxhash-libs                       x86_64 0.8.2-4.fc42               fedora      88.4 KiB
 xz-libs                           x86_64 1:5.6.3-2.fc42             fedora     218.4 KiB
 zig-srpm-macros                   noarch 1-3.fc41                   fedora       1.1 KiB
 zip                               x86_64 3.0-41.fc41                fedora     703.2 KiB
 zlib-ng-compat                    x86_64 2.2.2-1.fc42               fedora     134.0 KiB
 zstd                              x86_64 1.5.6-2.fc41               fedora       1.7 MiB
Installing groups:
 Buildsystem building group                                                              

Transaction Summary:
 Installing:       154 packages

Total size of inbound packages is 53 MiB. Need to download 53 MiB.
After this operation, 179 MiB extra will be used (install 179 MiB, remove 0 B).
[  1/154] bzip2-0:1.0.8-19.fc41.x86_64  100% | 280.9 KiB/s |  52.5 KiB |  00m00s
[  2/154] cpio-0:2.15-2.fc41.x86_64     100% |   1.9 MiB/s | 291.8 KiB |  00m00s
[  3/154] coreutils-0:9.5-11.fc42.x86_6 100% |   2.8 MiB/s |   1.1 MiB |  00m00s
[  4/154] diffutils-0:3.10-8.fc41.x86_6 100% |   4.3 MiB/s | 405.4 KiB |  00m00s
[  5/154] fedora-release-common-0:42-0. 100% | 547.1 KiB/s |  23.5 KiB |  00m00s
[  6/154] bash-0:5.2.37-1.fc42.x86_64   100% |   3.8 MiB/s |   1.8 MiB |  00m00s
[  7/154] glibc-minimal-langpack-0:2.40 100% |   2.2 MiB/s | 119.6 KiB |  00m00s
[  8/154] findutils-1:4.10.0-4.fc41.x86 100% |   5.5 MiB/s | 548.6 KiB |  00m00s
[  9/154] grep-0:3.11-9.fc41.x86_64     100% |   4.1 MiB/s | 299.8 KiB |  00m00s
[ 10/154] gzip-0:1.13-2.fc41.x86_64     100% |   2.9 MiB/s | 170.2 KiB |  00m00s
[ 11/154] patch-0:2.7.6-25.fc41.x86_64  100% |   2.5 MiB/s | 131.0 KiB |  00m00s
[ 12/154] redhat-rpm-config-0:296-1.fc4 100% |   1.7 MiB/s |  82.4 KiB |  00m00s
[ 13/154] info-0:7.1.1-2.fc42.x86_64    100% |   2.0 MiB/s | 183.2 KiB |  00m00s
[ 14/154] rpm-build-0:4.20.0-1.fc42.x86 100% |   1.7 MiB/s |  82.7 KiB |  00m00s
[ 15/154] sed-0:4.9-3.fc41.x86_64       100% |   4.3 MiB/s | 317.7 KiB |  00m00s
[ 16/154] unzip-0:6.0-65.fc42.x86_64    100% |   3.1 MiB/s | 184.5 KiB |  00m00s
[ 17/154] tar-2:1.35-4.fc41.x86_64      100% |   6.6 MiB/s | 860.7 KiB |  00m00s
[ 18/154] which-0:2.21-42.fc41.x86_64   100% | 989.5 KiB/s |  41.6 KiB |  00m00s
[ 19/154] shadow-utils-2:4.16.0-7.fc42. 100% |   7.5 MiB/s |   1.3 MiB |  00m00s
[ 20/154] xz-1:5.6.3-2.fc42.x86_64      100% |   5.0 MiB/s | 475.4 KiB |  00m00s
[ 21/154] gawk-0:5.3.0-4.fc41.x86_64    100% |   6.5 MiB/s |   1.1 MiB |  00m00s
[ 22/154] util-linux-0:2.40.2-8.fc42.x8 100% |   7.1 MiB/s |   1.2 MiB |  00m00s
[ 23/154] filesystem-0:3.18-29.fc42.x86 100% |   7.8 MiB/s |   1.1 MiB |  00m00s
[ 24/154] ncurses-libs-0:6.5-2.20240629 100% |   4.2 MiB/s | 334.0 KiB |  00m00s
[ 25/154] bzip2-libs-0:1.0.8-19.fc41.x8 100% | 913.6 KiB/s |  41.1 KiB |  00m00s
[ 26/154] gmp-1:6.3.0-2.fc41.x86_64     100% |   3.9 MiB/s | 318.0 KiB |  00m00s
[ 27/154] libacl-0:2.3.2-2.fc41.x86_64  100% | 583.3 KiB/s |  24.5 KiB |  00m00s
[ 28/154] glibc-0:2.40.9000-20.fc42.x86 100% |   9.0 MiB/s |   2.2 MiB |  00m00s
[ 29/154] libattr-0:2.5.2-4.fc41.x86_64 100% | 422.6 KiB/s |  18.2 KiB |  00m00s
[ 30/154] libcap-0:2.71-1.fc42.x86_64   100% |   1.7 MiB/s |  86.4 KiB |  00m00s
[ 31/154] libselinux-0:3.7-6.fc42.x86_6 100% |   1.9 MiB/s |  88.6 KiB |  00m00s
[ 32/154] coreutils-common-0:9.5-11.fc4 100% |   9.0 MiB/s |   2.1 MiB |  00m00s
[ 33/154] fedora-repos-0:42-0.3.noarch  100% | 218.8 KiB/s |   9.2 KiB |  00m00s
[ 34/154] systemd-libs-0:257~rc2-2.fc42 100% |   5.4 MiB/s | 813.0 KiB |  00m00s
[ 35/154] glibc-common-0:2.40.9000-20.f 100% |   4.1 MiB/s | 407.3 KiB |  00m00s
[ 36/154] ed-0:1.20.2-2.fc41.x86_64     100% |   1.5 MiB/s |  81.8 KiB |  00m00s
[ 37/154] pcre2-0:10.44-1.fc41.1.x86_64 100% |   3.2 MiB/s | 243.1 KiB |  00m00s
[ 38/154] ansible-srpm-macros-0:1-16.fc 100% | 483.1 KiB/s |  20.8 KiB |  00m00s
[ 39/154] build-reproducibility-srpm-ma 100% | 266.2 KiB/s |  11.2 KiB |  00m00s
[ 40/154] openssl-libs-1:3.2.2-8.fc42.x 100% |   7.3 MiB/s |   2.3 MiB |  00m00s
[ 41/154] dwz-0:0.15-8.fc42.x86_64      100% |   2.4 MiB/s | 139.1 KiB |  00m00s
[ 42/154] efi-srpm-macros-0:5-13.fc42.n 100% | 510.5 KiB/s |  22.5 KiB |  00m00s
[ 43/154] file-0:5.45-8.fc42.x86_64     100% |   1.0 MiB/s |  48.6 KiB |  00m00s
[ 44/154] fonts-srpm-macros-1:2.0.5-17. 100% | 642.1 KiB/s |  27.0 KiB |  00m00s
[ 45/154] forge-srpm-macros-0:0.4.0-1.f 100% | 481.9 KiB/s |  19.8 KiB |  00m00s
[ 46/154] fpc-srpm-macros-0:1.3-13.fc41 100% | 194.2 KiB/s |   8.0 KiB |  00m00s
[ 47/154] ghc-srpm-macros-0:1.9.2-1.fc4 100% | 222.4 KiB/s |   9.1 KiB |  00m00s
[ 48/154] gnat-srpm-macros-0:6-6.fc41.n 100% | 223.8 KiB/s |   9.0 KiB |  00m00s
[ 49/154] go-srpm-macros-0:3.6.0-3.fc41 100% | 621.1 KiB/s |  28.0 KiB |  00m00s
[ 50/154] kernel-srpm-macros-0:1.0-24.f 100% | 224.3 KiB/s |   9.9 KiB |  00m00s
[ 51/154] lua-srpm-macros-0:1-14.fc41.n 100% | 211.4 KiB/s |   8.9 KiB |  00m00s
[ 52/154] ocaml-srpm-macros-0:10-3.fc41 100% | 213.9 KiB/s |   9.2 KiB |  00m00s
[ 53/154] openblas-srpm-macros-0:2-18.f 100% | 188.2 KiB/s |   7.7 KiB |  00m00s
[ 54/154] package-notes-srpm-macros-0:0 100% | 228.6 KiB/s |   9.8 KiB |  00m00s
[ 55/154] perl-srpm-macros-0:1-56.fc41. 100% | 181.1 KiB/s |   8.5 KiB |  00m00s
[ 56/154] python-srpm-macros-0:3.13-3.f 100% | 551.7 KiB/s |  23.7 KiB |  00m00s
[ 57/154] pyproject-srpm-macros-0:1.16. 100% | 301.5 KiB/s |  13.9 KiB |  00m00s
[ 58/154] qt5-srpm-macros-0:5.15.15-1.f 100% | 202.2 KiB/s |   8.9 KiB |  00m00s
[ 59/154] qt6-srpm-macros-0:6.8.0-1.fc4 100% | 220.7 KiB/s |   9.0 KiB |  00m00s
[ 60/154] rpm-0:4.20.0-1.fc42.x86_64    100% |   7.1 MiB/s | 547.3 KiB |  00m00s
[ 61/154] zig-srpm-macros-0:1-3.fc41.no 100% | 188.9 KiB/s |   8.1 KiB |  00m00s
[ 62/154] rust-srpm-macros-0:26.3-3.fc4 100% | 281.4 KiB/s |  12.1 KiB |  00m00s
[ 63/154] zip-0:3.0-41.fc41.x86_64      100% |   4.7 MiB/s | 264.8 KiB |  00m00s
[ 64/154] debugedit-0:5.1-1.fc42.x86_64 100% |   1.5 MiB/s |  78.1 KiB |  00m00s
[ 65/154] elfutils-libelf-0:0.192-6.fc4 100% |   3.4 MiB/s | 207.5 KiB |  00m00s
[ 66/154] elfutils-0:0.192-6.fc42.x86_6 100% |   5.1 MiB/s | 550.7 KiB |  00m00s
[ 67/154] libarchive-0:3.7.7-1.fc42.x86 100% |   4.5 MiB/s | 413.9 KiB |  00m00s
[ 68/154] popt-0:1.19-7.fc41.x86_64     100% |   1.5 MiB/s |  65.9 KiB |  00m00s
[ 69/154] readline-0:8.2-11.fc42.x86_64 100% |   3.0 MiB/s | 213.4 KiB |  00m00s
[ 70/154] rpm-build-libs-0:4.20.0-1.fc4 100% |   1.8 MiB/s |  98.7 KiB |  00m00s
[ 71/154] rpm-libs-0:4.20.0-1.fc42.x86_ 100% |   5.8 MiB/s | 309.5 KiB |  00m00s
[ 72/154] libeconf-0:0.7.4-3.fc42.x86_6 100% | 805.5 KiB/s |  34.6 KiB |  00m00s
[ 73/154] audit-libs-0:4.0.2-1.fc41.x86 100% |   2.1 MiB/s | 126.2 KiB |  00m00s
[ 74/154] zstd-0:1.5.6-2.fc41.x86_64    100% |   4.8 MiB/s | 481.5 KiB |  00m00s
[ 75/154] libsemanage-0:3.7-3.fc42.x86_ 100% |   2.7 MiB/s | 117.2 KiB |  00m00s
[ 76/154] libxcrypt-0:4.4.36-10.fc42.x8 100% |   2.2 MiB/s | 119.0 KiB |  00m00s
[ 77/154] pam-libs-0:1.7.0-3.fc42.x86_6 100% |   1.2 MiB/s |  58.0 KiB |  00m00s
[ 78/154] setup-0:2.15.0-5.fc41.noarch  100% |   3.4 MiB/s | 154.4 KiB |  00m00s
[ 79/154] xz-libs-1:5.6.3-2.fc42.x86_64 100% |   2.1 MiB/s | 111.9 KiB |  00m00s
[ 80/154] libblkid-0:2.40.2-8.fc42.x86_ 100% |   2.7 MiB/s | 125.0 KiB |  00m00s
[ 81/154] mpfr-0:4.2.1-5.fc41.x86_64    100% |   4.0 MiB/s | 346.3 KiB |  00m00s
[ 82/154] libcap-ng-0:0.8.5-3.fc41.x86_ 100% | 757.3 KiB/s |  32.6 KiB |  00m00s
[ 83/154] libfdisk-0:2.40.2-8.fc42.x86_ 100% |   3.4 MiB/s | 159.6 KiB |  00m00s
[ 84/154] libsmartcols-0:2.40.2-8.fc42. 100% |   1.5 MiB/s |  83.9 KiB |  00m00s
[ 85/154] libmount-0:2.40.2-8.fc42.x86_ 100% |   2.4 MiB/s | 156.1 KiB |  00m00s
[ 86/154] libuuid-0:2.40.2-8.fc42.x86_6 100% | 690.0 KiB/s |  29.0 KiB |  00m00s
[ 87/154] zlib-ng-compat-0:2.2.2-1.fc42 100% |   1.4 MiB/s |  76.9 KiB |  00m00s
[ 88/154] basesystem-0:11-21.fc41.noarc 100% | 134.0 KiB/s |   7.4 KiB |  00m00s
[ 89/154] glibc-gconv-extra-0:2.40.9000 100% |  13.2 MiB/s |   1.7 MiB |  00m00s
[ 90/154] util-linux-core-0:2.40.2-8.fc 100% |   3.8 MiB/s | 537.3 KiB |  00m00s
[ 91/154] libgcc-0:14.2.1-6.fc42.x86_64 100% |   2.3 MiB/s | 135.2 KiB |  00m00s
[ 92/154] ncurses-base-0:6.5-2.20240629 100% |   1.9 MiB/s |  88.4 KiB |  00m00s
[ 93/154] crypto-policies-0:20241106-1. 100% |   2.0 MiB/s |  98.0 KiB |  00m00s
[ 94/154] libsepol-0:3.7-3.fc42.x86_64  100% |   3.5 MiB/s | 342.2 KiB |  00m00s
[ 95/154] fedora-repos-rawhide-0:42-0.3 100% | 199.2 KiB/s |   8.8 KiB |  00m00s
[ 96/154] fedora-gpg-keys-0:42-0.3.noar 100% |   2.5 MiB/s | 133.6 KiB |  00m00s
[ 97/154] ca-certificates-0:2024.2.69_v 100% |   5.2 MiB/s | 944.5 KiB |  00m00s
[ 98/154] pcre2-syntax-0:10.44-1.fc41.1 100% |   1.9 MiB/s | 149.9 KiB |  00m00s
[ 99/154] add-determinism-0:0.4.3-1.fc4 100% |   7.1 MiB/s | 904.4 KiB |  00m00s
[100/154] curl-0:8.10.1-2.fc42.x86_64   100% |   2.5 MiB/s | 221.3 KiB |  00m00s
[101/154] elfutils-libs-0:0.192-6.fc42. 100% |   4.4 MiB/s | 263.4 KiB |  00m00s
[102/154] elfutils-debuginfod-client-0: 100% | 958.5 KiB/s |  46.0 KiB |  00m00s
[103/154] file-libs-0:5.45-8.fc42.x86_6 100% |   5.0 MiB/s | 763.6 KiB |  00m00s
[104/154] libstdc++-0:14.2.1-6.fc42.x86 100% |   8.7 MiB/s | 890.1 KiB |  00m00s
[105/154] libzstd-0:1.5.6-2.fc41.x86_64 100% |   2.9 MiB/s | 310.3 KiB |  00m00s
[106/154] lz4-libs-0:1.10.0-1.fc41.x86_ 100% |   1.5 MiB/s |  70.7 KiB |  00m00s
[107/154] libxml2-0:2.12.8-2.fc41.x86_6 100% |   4.8 MiB/s | 687.3 KiB |  00m00s
[108/154] lua-libs-0:5.4.7-1.fc42.x86_6 100% |   2.7 MiB/s | 132.1 KiB |  00m00s
[109/154] libgomp-0:14.2.1-6.fc42.x86_6 100% |   3.8 MiB/s | 356.9 KiB |  00m00s
[110/154] elfutils-default-yama-scope-0 100% | 293.5 KiB/s |  12.3 KiB |  00m00s
[111/154] sqlite-libs-0:3.47.0-1.fc42.x 100% |   8.9 MiB/s | 723.2 KiB |  00m00s
[112/154] json-c-0:0.18-1.fc42.x86_64   100% | 887.9 KiB/s |  44.4 KiB |  00m00s
[113/154] authselect-libs-0:1.5.0-8.fc4 100% |   3.3 MiB/s | 218.0 KiB |  00m00s
[114/154] rpm-sequoia-0:1.7.0-2.fc41.x8 100% |   5.2 MiB/s | 892.5 KiB |  00m00s
[115/154] authselect-0:1.5.0-8.fc42.x86 100% |   2.5 MiB/s | 145.8 KiB |  00m00s
[116/154] gdbm-libs-1:1.23-7.fc41.x86_6 100% |   1.1 MiB/s |  56.3 KiB |  00m00s
[117/154] libnsl2-0:2.0.1-2.fc41.x86_64 100% | 658.0 KiB/s |  29.6 KiB |  00m00s
[118/154] pam-0:1.7.0-3.fc42.x86_64     100% |   3.9 MiB/s | 554.3 KiB |  00m00s
[119/154] libpwquality-0:1.4.5-11.fc41. 100% |   1.9 MiB/s | 119.1 KiB |  00m00s
[120/154] libtirpc-0:1.3.6-1.fc42.x86_6 100% |   1.9 MiB/s |  94.9 KiB |  00m00s
[121/154] cracklib-0:2.9.11-6.fc41.x86_ 100% |   1.5 MiB/s |  92.0 KiB |  00m00s
[122/154] libcom_err-0:1.47.1-6.fc42.x8 100% | 618.0 KiB/s |  26.6 KiB |  00m00s
[123/154] keyutils-libs-0:1.6.3-4.fc41. 100% | 673.4 KiB/s |  31.6 KiB |  00m00s
[124/154] libverto-0:0.3.2-9.fc41.x86_6 100% | 470.3 KiB/s |  20.7 KiB |  00m00s
[125/154] alternatives-0:1.30-1.fc41.x8 100% | 923.1 KiB/s |  42.5 KiB |  00m00s
[126/154] krb5-libs-0:1.21.3-3.fc42.x86 100% |   4.4 MiB/s | 760.4 KiB |  00m00s
[127/154] jansson-0:2.14-1.fc42.x86_64  100% | 888.1 KiB/s |  45.3 KiB |  00m00s
[128/154] pkgconf-pkg-config-0:2.3.0-1. 100% | 200.2 KiB/s |  10.0 KiB |  00m00s
[129/154] pkgconf-m4-0:2.3.0-1.fc42.noa 100% | 325.5 KiB/s |  14.3 KiB |  00m00s
[130/154] pkgconf-0:2.3.0-1.fc42.x86_64 100% |   1.0 MiB/s |  45.2 KiB |  00m00s
[131/154] libpkgconf-0:2.3.0-1.fc42.x86 100% | 855.3 KiB/s |  38.5 KiB |  00m00s
[132/154] gdbm-1:1.23-7.fc41.x86_64     100% |   2.2 MiB/s | 151.8 KiB |  00m00s
[133/154] libffi-0:3.4.6-3.fc42.x86_64  100% | 975.4 KiB/s |  40.0 KiB |  00m00s
[134/154] libtasn1-0:4.19.0-9.fc41.x86_ 100% |   1.7 MiB/s |  74.2 KiB |  00m00s
[135/154] p11-kit-0:0.25.5-4.fc42.x86_6 100% |   4.8 MiB/s | 492.0 KiB |  00m00s
[136/154] p11-kit-trust-0:0.25.5-4.fc42 100% |   2.7 MiB/s | 133.2 KiB |  00m00s
[137/154] fedora-release-0:42-0.8.noarc 100% | 302.2 KiB/s |  12.7 KiB |  00m00s
[138/154] xxhash-libs-0:0.8.2-4.fc42.x8 100% | 817.2 KiB/s |  36.8 KiB |  00m00s
[139/154] fedora-release-identity-basic 100% | 313.9 KiB/s |  13.5 KiB |  00m00s
[140/154] libcurl-0:8.10.1-2.fc42.x86_6 100% |   4.0 MiB/s | 371.3 KiB |  00m00s
[141/154] libbrotli-0:1.1.0-5.fc41.x86_ 100% |   3.9 MiB/s | 340.5 KiB |  00m00s
[142/154] gdb-minimal-0:15.2-3.fc42.x86 100% |  12.8 MiB/s |   4.3 MiB |  00m00s
[143/154] libnghttp2-0:1.64.0-1.fc42.x8 100% |   1.7 MiB/s |  77.4 KiB |  00m00s
[144/154] libpsl-0:0.21.5-4.fc41.x86_64 100% |   1.5 MiB/s |  64.1 KiB |  00m00s
[145/154] libidn2-0:2.3.7-2.fc41.x86_64 100% | 870.5 KiB/s | 118.4 KiB |  00m00s
[146/154] libssh-0:0.11.1-1.fc42.x86_64 100% |   3.8 MiB/s | 231.9 KiB |  00m00s
[147/154] openldap-0:2.6.8-5.fc41.x86_6 100% |   2.8 MiB/s | 255.6 KiB |  00m00s
[148/154] libunistring-0:1.1-8.fc41.x86 100% |   5.8 MiB/s | 544.8 KiB |  00m00s
[149/154] publicsuffix-list-dafsa-0:202 100% |   1.1 MiB/s |  58.3 KiB |  00m00s
[150/154] binutils-0:2.43.50-8.fc42.x86 100% |   6.0 MiB/s |   5.8 MiB |  00m01s
[151/154] libssh-config-0:0.11.1-1.fc42 100% | 223.5 KiB/s |   9.4 KiB |  00m00s
[152/154] libtool-ltdl-0:2.4.7-12.fc41. 100% | 790.9 KiB/s |  35.6 KiB |  00m00s
[153/154] libevent-0:2.1.12-14.fc41.x86 100% |   3.1 MiB/s | 257.5 KiB |  00m00s
[154/154] cyrus-sasl-lib-0:2.1.28-27.fc 100% |   5.2 MiB/s | 794.9 KiB |  00m00s
--------------------------------------------------------------------------------
[154/154] Total                         100% |  11.8 MiB/s |  52.7 MiB |  00m04s
Running transaction
Importing PGP key 0x105EF944:
 UserID     : "Fedora (42) <fedora-42-primary@fedoraproject.org>"
 Fingerprint: B0F4950458F69E1150C6C5EDC8AC4916105EF944
 From       : file:///usr/share/distribution-gpg-keys/fedora/RPM-GPG-KEY-fedora-42-primary
The key was successfully imported.
Importing PGP key 0x105EF944:
 UserID     : "Fedora (42) <fedora-42-primary@fedoraproject.org>"
 Fingerprint: B0F4950458F69E1150C6C5EDC8AC4916105EF944
 From       : file:///usr/share/distribution-gpg-keys/fedora/RPM-GPG-KEY-fedora-42-primary
The key was successfully imported.
Importing PGP key 0xE99D6AD1:
 UserID     : "Fedora (41) <fedora-41-primary@fedoraproject.org>"
 Fingerprint: 466CF2D8B60BC3057AA9453ED0622462E99D6AD1
 From       : file:///usr/share/distribution-gpg-keys/fedora/RPM-GPG-KEY-fedora-41-primary
The key was successfully imported.
Importing PGP key 0x31645531:
 UserID     : "Fedora (43) <fedora-43-primary@fedoraproject.org>"
 Fingerprint: C6E7F081CF80E13146676E88829B606631645531
 From       : file:///usr/share/distribution-gpg-keys/fedora/RPM-GPG-KEY-fedora-43-primary
The key was successfully imported.
[  1/156] Verify package files          100% | 751.0   B/s | 154.0   B |  00m00s
[  2/156] Prepare transaction           100% |   1.8 KiB/s | 154.0   B |  00m00s
[  3/156] Installing libgcc-0:14.2.1-6. 100% | 132.9 MiB/s | 272.3 KiB |  00m00s
[  4/156] Installing libssh-config-0:0. 100% |   0.0   B/s | 816.0   B |  00m00s
[  5/156] Installing publicsuffix-list- 100% |  66.7 MiB/s |  68.3 KiB |  00m00s
[  6/156] Installing fedora-release-ide 100% |   0.0   B/s | 976.0   B |  00m00s
[  7/156] Installing fedora-repos-rawhi 100% |   2.4 MiB/s |   2.4 KiB |  00m00s
[  8/156] Installing fedora-gpg-keys-0: 100% |  18.7 MiB/s | 172.2 KiB |  00m00s
[  9/156] Installing fedora-repos-0:42- 100% |   0.0   B/s |   5.7 KiB |  00m00s
[ 10/156] Installing fedora-release-com 100% |  11.7 MiB/s |  24.0 KiB |  00m00s
[ 11/156] Installing fedora-release-0:4 100% |   0.0   B/s | 124.0   B |  00m00s
[ 12/156] Installing setup-0:2.15.0-5.f 100% |  44.3 MiB/s | 726.1 KiB |  00m00s
>>> [RPM] /etc/hosts created as /etc/hosts.rpmnew
[ 13/156] Installing filesystem-0:3.18- 100% |   1.6 MiB/s | 212.6 KiB |  00m00s
[ 14/156] Installing basesystem-0:11-21 100% |   0.0   B/s | 124.0   B |  00m00s
[ 15/156] Installing pkgconf-m4-0:2.3.0 100% |   0.0   B/s |  14.8 KiB |  00m00s
[ 16/156] Installing pcre2-syntax-0:10. 100% | 124.1 MiB/s | 254.1 KiB |  00m00s
[ 17/156] Installing ncurses-base-0:6.5 100% |  34.3 MiB/s | 351.7 KiB |  00m00s
[ 18/156] Installing glibc-minimal-lang 100% |   0.0   B/s | 124.0   B |  00m00s
[ 19/156] Installing ncurses-libs-0:6.5 100% | 137.0 MiB/s | 981.8 KiB |  00m00s
[ 20/156] Installing glibc-0:2.40.9000- 100% | 197.8 MiB/s |   6.7 MiB |  00m00s
[ 21/156] Installing bash-0:5.2.37-1.fc 100% | 272.3 MiB/s |   8.2 MiB |  00m00s
[ 22/156] Installing glibc-common-0:2.4 100% | 117.7 MiB/s |   1.1 MiB |  00m00s
[ 23/156] Installing glibc-gconv-extra- 100% | 150.7 MiB/s |   8.1 MiB |  00m00s
[ 24/156] Installing zlib-ng-compat-0:2 100% | 131.7 MiB/s | 134.9 KiB |  00m00s
[ 25/156] Installing bzip2-libs-0:1.0.8 100% |  79.9 MiB/s |  81.8 KiB |  00m00s
[ 26/156] Installing xz-libs-1:5.6.3-2. 100% | 214.3 MiB/s | 219.5 KiB |  00m00s
[ 27/156] Installing popt-0:1.19-7.fc41 100% |  28.0 MiB/s | 143.5 KiB |  00m00s
[ 28/156] Installing readline-0:8.2-11. 100% | 161.2 MiB/s | 495.3 KiB |  00m00s
[ 29/156] Installing libuuid-0:2.40.2-8 100% |  41.5 MiB/s |  42.5 KiB |  00m00s
[ 30/156] Installing libblkid-0:2.40.2- 100% | 128.7 MiB/s | 263.6 KiB |  00m00s
[ 31/156] Installing gmp-1:6.3.0-2.fc41 100% | 198.6 MiB/s | 813.7 KiB |  00m00s
[ 32/156] Installing libattr-0:2.5.2-4. 100% |  28.8 MiB/s |  29.5 KiB |  00m00s
[ 33/156] Installing libacl-0:2.3.2-2.f 100% |  39.8 MiB/s |  40.7 KiB |  00m00s
[ 34/156] Installing libxcrypt-0:4.4.36 100% |  88.0 MiB/s | 270.4 KiB |  00m00s
[ 35/156] Installing libstdc++-0:14.2.1 100% | 250.9 MiB/s |   2.8 MiB |  00m00s
[ 36/156] Installing libzstd-0:1.5.6-2. 100% | 259.5 MiB/s | 797.2 KiB |  00m00s
[ 37/156] Installing elfutils-libelf-0: 100% | 233.3 MiB/s |   1.2 MiB |  00m00s
[ 38/156] Installing libeconf-0:0.7.4-3 100% |  65.8 MiB/s |  67.4 KiB |  00m00s
[ 39/156] Installing gdbm-libs-1:1.23-7 100% | 120.7 MiB/s | 123.6 KiB |  00m00s
[ 40/156] Installing dwz-0:0.15-8.fc42. 100% | 146.8 MiB/s | 300.6 KiB |  00m00s
[ 41/156] Installing mpfr-0:4.2.1-5.fc4 100% | 203.5 MiB/s | 833.7 KiB |  00m00s
[ 42/156] Installing gawk-0:5.3.0-4.fc4 100% | 173.2 MiB/s |   1.7 MiB |  00m00s
[ 43/156] Installing unzip-0:6.0-65.fc4 100% | 130.8 MiB/s | 401.7 KiB |  00m00s
[ 44/156] Installing file-libs-0:5.45-8 100% | 496.9 MiB/s |   9.9 MiB |  00m00s
[ 45/156] Installing file-0:5.45-8.fc42 100% |   8.6 MiB/s | 105.2 KiB |  00m00s
[ 46/156] Installing crypto-policies-0: 100% |  14.5 MiB/s | 163.3 KiB |  00m00s
[ 47/156] Installing pcre2-0:10.44-1.fc 100% | 213.2 MiB/s | 654.9 KiB |  00m00s
[ 48/156] Installing grep-0:3.11-9.fc41 100% | 111.5 MiB/s |   1.0 MiB |  00m00s
[ 49/156] Installing xz-1:5.6.3-2.fc42. 100% | 123.7 MiB/s |   1.2 MiB |  00m00s
[ 50/156] Installing libcap-ng-0:0.8.5- 100% |  69.4 MiB/s |  71.0 KiB |  00m00s
[ 51/156] Installing audit-libs-0:4.0.2 100% | 162.8 MiB/s | 333.4 KiB |  00m00s
[ 52/156] Installing pam-libs-0:1.7.0-3 100% |  69.2 MiB/s | 141.8 KiB |  00m00s
[ 53/156] Installing libcap-0:2.71-1.fc 100% |  70.3 MiB/s | 215.8 KiB |  00m00s
[ 54/156] Installing systemd-libs-0:257 100% | 253.1 MiB/s |   2.3 MiB |  00m00s
[ 55/156] Installing libsmartcols-0:2.4 100% | 177.1 MiB/s | 181.4 KiB |  00m00s
[ 56/156] Installing libsepol-0:3.7-3.f 100% | 266.6 MiB/s | 819.0 KiB |  00m00s
[ 57/156] Installing libselinux-0:3.7-6 100% |  89.0 MiB/s | 182.3 KiB |  00m00s
[ 58/156] Installing sed-0:4.9-3.fc41.x 100% | 121.3 MiB/s | 869.7 KiB |  00m00s
[ 59/156] Installing findutils-1:4.10.0 100% | 185.8 MiB/s |   1.9 MiB |  00m00s
[ 60/156] Installing libmount-0:2.40.2- 100% | 174.3 MiB/s | 356.9 KiB |  00m00s
[ 61/156] Installing lz4-libs-0:1.10.0- 100% | 143.1 MiB/s | 146.6 KiB |  00m00s
[ 62/156] Installing lua-libs-0:5.4.7-1 100% | 139.7 MiB/s | 286.2 KiB |  00m00s
[ 63/156] Installing libcom_err-0:1.47. 100% |  66.7 MiB/s |  68.3 KiB |  00m00s
[ 64/156] Installing alternatives-0:1.3 100% |  66.3 MiB/s |  67.9 KiB |  00m00s
[ 65/156] Installing libffi-0:3.4.6-3.f 100% |  85.7 MiB/s |  87.8 KiB |  00m00s
[ 66/156] Installing libtasn1-0:4.19.0- 100% |  86.7 MiB/s | 177.5 KiB |  00m00s
[ 67/156] Installing p11-kit-0:0.25.5-4 100% | 147.7 MiB/s |   2.2 MiB |  00m00s
[ 68/156] Installing libunistring-0:1.1 100% | 247.2 MiB/s |   1.7 MiB |  00m00s
[ 69/156] Installing libidn2-0:2.3.7-2. 100% |  65.4 MiB/s | 335.1 KiB |  00m00s
[ 70/156] Installing libpsl-0:0.21.5-4. 100% |  79.7 MiB/s |  81.7 KiB |  00m00s
[ 71/156] Installing p11-kit-trust-0:0. 100% |  28.3 MiB/s | 405.5 KiB |  00m00s
[ 72/156] Installing zstd-0:1.5.6-2.fc4 100% | 241.6 MiB/s |   1.7 MiB |  00m00s
[ 73/156] Installing util-linux-core-0: 100% | 127.3 MiB/s |   1.5 MiB |  00m00s
[ 74/156] Installing tar-2:1.35-4.fc41. 100% | 211.3 MiB/s |   3.0 MiB |  00m00s
[ 75/156] Installing libsemanage-0:3.7- 100% |  97.6 MiB/s | 299.8 KiB |  00m00s
[ 76/156] Installing shadow-utils-2:4.1 100% | 120.4 MiB/s |   4.1 MiB |  00m00s
[ 77/156] Installing zip-0:3.0-41.fc41. 100% | 172.6 MiB/s | 707.1 KiB |  00m00s
[ 78/156] Installing gdbm-1:1.23-7.fc41 100% | 113.7 MiB/s | 465.8 KiB |  00m00s
[ 79/156] Installing cyrus-sasl-lib-0:2 100% | 256.2 MiB/s |   2.3 MiB |  00m00s
[ 80/156] Installing libfdisk-0:2.40.2- 100% | 177.7 MiB/s | 364.0 KiB |  00m00s
[ 81/156] Installing libxml2-0:2.12.8-2 100% | 214.0 MiB/s |   1.7 MiB |  00m00s
[ 82/156] Installing bzip2-0:1.0.8-19.f 100% |  48.9 MiB/s | 100.2 KiB |  00m00s
[ 83/156] Installing add-determinism-0: 100% | 270.1 MiB/s |   2.4 MiB |  00m00s
[ 84/156] Installing build-reproducibil 100% |   0.0   B/s |   1.0 KiB |  00m00s
[ 85/156] Installing sqlite-libs-0:3.47 100% | 242.2 MiB/s |   1.5 MiB |  00m00s
[ 86/156] Installing ed-0:1.20.2-2.fc41 100% |  72.8 MiB/s | 149.2 KiB |  00m00s
[ 87/156] Installing patch-0:2.7.6-25.f 100% |  52.4 MiB/s | 268.2 KiB |  00m00s
[ 88/156] Installing elfutils-default-y 100% | 204.3 KiB/s |   2.0 KiB |  00m00s
[ 89/156] Installing elfutils-libs-0:0. 100% | 164.7 MiB/s | 674.7 KiB |  00m00s
[ 90/156] Installing cpio-0:2.15-2.fc41 100% | 137.5 MiB/s |   1.1 MiB |  00m00s
[ 91/156] Installing diffutils-0:3.10-8 100% | 159.0 MiB/s |   1.6 MiB |  00m00s
[ 92/156] Installing libgomp-0:14.2.1-6 100% | 254.5 MiB/s | 521.2 KiB |  00m00s
[ 93/156] Installing json-c-0:0.18-1.fc 100% |  82.6 MiB/s |  84.6 KiB |  00m00s
[ 94/156] Installing keyutils-libs-0:1. 100% |  54.5 MiB/s |  55.8 KiB |  00m00s
[ 95/156] Installing libverto-0:0.3.2-9 100% |  30.5 MiB/s |  31.3 KiB |  00m00s
[ 96/156] Installing jansson-0:2.14-1.f 100% |  92.3 MiB/s |  94.5 KiB |  00m00s
[ 97/156] Installing libpkgconf-0:2.3.0 100% |  77.5 MiB/s |  79.3 KiB |  00m00s
[ 98/156] Installing pkgconf-0:2.3.0-1. 100% |  44.5 MiB/s |  91.1 KiB |  00m00s
[ 99/156] Installing pkgconf-pkg-config 100% |   1.7 MiB/s |   1.8 KiB |  00m00s
[100/156] Installing xxhash-libs-0:0.8. 100% |  87.7 MiB/s |  89.8 KiB |  00m00s
[101/156] Installing libbrotli-0:1.1.0- 100% | 205.0 MiB/s | 839.9 KiB |  00m00s
[102/156] Installing libnghttp2-0:1.64. 100% | 171.5 MiB/s | 175.6 KiB |  00m00s
[103/156] Installing libtool-ltdl-0:2.4 100% |  65.7 MiB/s |  67.3 KiB |  00m00s
[104/156] Installing rust-srpm-macros-0 100% |   0.0   B/s |   5.6 KiB |  00m00s
[105/156] Installing qt6-srpm-macros-0: 100% |   0.0   B/s | 732.0   B |  00m00s
[106/156] Installing qt5-srpm-macros-0: 100% |   0.0   B/s | 776.0   B |  00m00s
[107/156] Installing perl-srpm-macros-0 100% |   0.0   B/s |   1.1 KiB |  00m00s
[108/156] Installing package-notes-srpm 100% |   0.0   B/s |   2.0 KiB |  00m00s
[109/156] Installing openblas-srpm-macr 100% |   0.0   B/s | 392.0   B |  00m00s
[110/156] Installing ocaml-srpm-macros- 100% |   0.0   B/s |   2.2 KiB |  00m00s
[111/156] Installing kernel-srpm-macros 100% |   0.0   B/s |   2.3 KiB |  00m00s
[112/156] Installing gnat-srpm-macros-0 100% |   0.0   B/s |   1.3 KiB |  00m00s
[113/156] Installing ghc-srpm-macros-0: 100% |   0.0   B/s |   1.0 KiB |  00m00s
[114/156] Installing fpc-srpm-macros-0: 100% |   0.0   B/s | 420.0   B |  00m00s
[115/156] Installing ansible-srpm-macro 100% |  35.4 MiB/s |  36.2 KiB |  00m00s
[116/156] Installing coreutils-common-0 100% | 243.3 MiB/s |  11.2 MiB |  00m00s
[117/156] Installing openssl-libs-1:3.2 100% | 313.1 MiB/s |   7.8 MiB |  00m00s
[118/156] Installing coreutils-0:9.5-11 100% | 142.1 MiB/s |   5.4 MiB |  00m00s
[119/156] Installing ca-certificates-0: 100% |   1.2 MiB/s |   2.4 MiB |  00m02s
[120/156] Installing krb5-libs-0:1.21.3 100% | 177.3 MiB/s |   2.3 MiB |  00m00s
[121/156] Installing libarchive-0:3.7.7 100% | 228.1 MiB/s | 934.2 KiB |  00m00s
[122/156] Installing libtirpc-0:1.3.6-1 100% | 101.2 MiB/s | 207.3 KiB |  00m00s
[123/156] Installing gzip-0:1.13-2.fc41 100% |  96.3 MiB/s | 394.6 KiB |  00m00s
[124/156] Installing authselect-libs-0: 100% |  81.8 MiB/s | 837.2 KiB |  00m00s
[125/156] Installing cracklib-0:2.9.11- 100% |  34.9 MiB/s | 250.3 KiB |  00m00s
[126/156] Installing libpwquality-0:1.4 100% |  52.5 MiB/s | 430.1 KiB |  00m00s
[127/156] Installing libnsl2-0:2.0.1-2. 100% |  28.8 MiB/s |  59.1 KiB |  00m00s
[128/156] Installing pam-0:1.7.0-3.fc42 100% |  75.7 MiB/s |   1.9 MiB |  00m00s
[129/156] Installing libssh-0:0.11.1-1. 100% | 186.1 MiB/s | 571.7 KiB |  00m00s
[130/156] Installing rpm-sequoia-0:1.7. 100% | 263.0 MiB/s |   2.4 MiB |  00m00s
[131/156] Installing rpm-libs-0:4.20.0- 100% | 177.7 MiB/s | 727.7 KiB |  00m00s
[132/156] Installing rpm-build-libs-0:4 100% | 101.3 MiB/s | 207.5 KiB |  00m00s
[133/156] Installing libevent-0:2.1.12- 100% | 175.7 MiB/s | 899.5 KiB |  00m00s
[134/156] Installing openldap-0:2.6.8-5 100% | 158.2 MiB/s | 648.0 KiB |  00m00s
[135/156] Installing libcurl-0:8.10.1-2 100% | 205.0 MiB/s | 839.5 KiB |  00m00s
[136/156] Installing elfutils-debuginfo 100% |  40.3 MiB/s |  82.6 KiB |  00m00s
[137/156] Installing elfutils-0:0.192-6 100% | 220.4 MiB/s |   2.6 MiB |  00m00s
[138/156] Installing binutils-0:2.43.50 100% | 262.4 MiB/s |  25.5 MiB |  00m00s
[139/156] Installing gdb-minimal-0:15.2 100% | 302.2 MiB/s |  13.0 MiB |  00m00s
[140/156] Installing debugedit-0:5.1-1. 100% |  95.4 MiB/s | 195.4 KiB |  00m00s
[141/156] Installing curl-0:8.10.1-2.fc 100% |  24.7 MiB/s | 455.8 KiB |  00m00s
[142/156] Installing rpm-0:4.20.0-1.fc4 100% |  86.4 MiB/s |   2.5 MiB |  00m00s
[143/156] Installing efi-srpm-macros-0: 100% |  40.2 MiB/s |  41.2 KiB |  00m00s
[144/156] Installing lua-srpm-macros-0: 100% |   0.0   B/s |   1.9 KiB |  00m00s
[145/156] Installing zig-srpm-macros-0: 100% |   0.0   B/s |   1.7 KiB |  00m00s
[146/156] Installing fonts-srpm-macros- 100% |  55.7 MiB/s |  57.0 KiB |  00m00s
[147/156] Installing forge-srpm-macros- 100% |  39.3 MiB/s |  40.3 KiB |  00m00s
[148/156] Installing go-srpm-macros-0:3 100% |  60.5 MiB/s |  62.0 KiB |  00m00s
[149/156] Installing python-srpm-macros 100% |  50.9 MiB/s |  52.2 KiB |  00m00s
[150/156] Installing redhat-rpm-config- 100% |  47.2 MiB/s | 193.2 KiB |  00m00s
[151/156] Installing rpm-build-0:4.20.0 100% |  49.5 MiB/s | 202.9 KiB |  00m00s
[152/156] Installing pyproject-srpm-mac 100% |   1.2 MiB/s |   2.5 KiB |  00m00s
[153/156] Installing util-linux-0:2.40. 100% |  90.6 MiB/s |   3.7 MiB |  00m00s
[154/156] Installing authselect-0:1.5.0 100% |  39.5 MiB/s | 161.9 KiB |  00m00s
[155/156] Installing which-0:2.21-42.fc 100% |  40.2 MiB/s |  82.4 KiB |  00m00s
[156/156] Installing info-0:7.1.1-2.fc4 100% | 138.3 KiB/s | 362.2 KiB |  00m03s
Complete!
Finish: installing minimal buildroot with dnf5
Start: creating root cache
Finish: creating root cache
Finish: chroot init
INFO: Installed packages:
INFO: add-determinism-0.4.3-1.fc42.x86_64
alternatives-1.30-1.fc41.x86_64
ansible-srpm-macros-1-16.fc41.noarch
audit-libs-4.0.2-1.fc41.x86_64
authselect-1.5.0-8.fc42.x86_64
authselect-libs-1.5.0-8.fc42.x86_64
basesystem-11-21.fc41.noarch
bash-5.2.37-1.fc42.x86_64
binutils-2.43.50-8.fc42.x86_64
build-reproducibility-srpm-macros-0.4.3-1.fc42.noarch
bzip2-1.0.8-19.fc41.x86_64
bzip2-libs-1.0.8-19.fc41.x86_64
ca-certificates-2024.2.69_v8.0.401-2.fc42.noarch
coreutils-9.5-11.fc42.x86_64
coreutils-common-9.5-11.fc42.x86_64
cpio-2.15-2.fc41.x86_64
cracklib-2.9.11-6.fc41.x86_64
crypto-policies-20241106-1.git35892de.fc42.noarch
curl-8.10.1-2.fc42.x86_64
cyrus-sasl-lib-2.1.28-27.fc41.x86_64
debugedit-5.1-1.fc42.x86_64
diffutils-3.10-8.fc41.x86_64
dwz-0.15-8.fc42.x86_64
ed-1.20.2-2.fc41.x86_64
efi-srpm-macros-5-13.fc42.noarch
elfutils-0.192-6.fc42.x86_64
elfutils-debuginfod-client-0.192-6.fc42.x86_64
elfutils-default-yama-scope-0.192-6.fc42.noarch
elfutils-libelf-0.192-6.fc42.x86_64
elfutils-libs-0.192-6.fc42.x86_64
fedora-gpg-keys-42-0.3.noarch
fedora-release-42-0.8.noarch
fedora-release-common-42-0.8.noarch
fedora-release-identity-basic-42-0.8.noarch
fedora-repos-42-0.3.noarch
fedora-repos-rawhide-42-0.3.noarch
file-5.45-8.fc42.x86_64
file-libs-5.45-8.fc42.x86_64
filesystem-3.18-29.fc42.x86_64
findutils-4.10.0-4.fc41.x86_64
fonts-srpm-macros-2.0.5-17.fc41.noarch
forge-srpm-macros-0.4.0-1.fc42.noarch
fpc-srpm-macros-1.3-13.fc41.noarch
gawk-5.3.0-4.fc41.x86_64
gdb-minimal-15.2-3.fc42.x86_64
gdbm-1.23-7.fc41.x86_64
gdbm-libs-1.23-7.fc41.x86_64
ghc-srpm-macros-1.9.2-1.fc42.noarch
glibc-2.40.9000-20.fc42.x86_64
glibc-common-2.40.9000-20.fc42.x86_64
glibc-gconv-extra-2.40.9000-20.fc42.x86_64
glibc-minimal-langpack-2.40.9000-20.fc42.x86_64
gmp-6.3.0-2.fc41.x86_64
gnat-srpm-macros-6-6.fc41.noarch
go-srpm-macros-3.6.0-3.fc41.noarch
gpg-pubkey-105ef944-65ca83d1
gpg-pubkey-31645531-66b6dccf
gpg-pubkey-e99d6ad1-64d2612c
grep-3.11-9.fc41.x86_64
gzip-1.13-2.fc41.x86_64
info-7.1.1-2.fc42.x86_64
jansson-2.14-1.fc42.x86_64
json-c-0.18-1.fc42.x86_64
kernel-srpm-macros-1.0-24.fc41.noarch
keyutils-libs-1.6.3-4.fc41.x86_64
krb5-libs-1.21.3-3.fc42.x86_64
libacl-2.3.2-2.fc41.x86_64
libarchive-3.7.7-1.fc42.x86_64
libattr-2.5.2-4.fc41.x86_64
libblkid-2.40.2-8.fc42.x86_64
libbrotli-1.1.0-5.fc41.x86_64
libcap-2.71-1.fc42.x86_64
libcap-ng-0.8.5-3.fc41.x86_64
libcom_err-1.47.1-6.fc42.x86_64
libcurl-8.10.1-2.fc42.x86_64
libeconf-0.7.4-3.fc42.x86_64
libevent-2.1.12-14.fc41.x86_64
libfdisk-2.40.2-8.fc42.x86_64
libffi-3.4.6-3.fc42.x86_64
libgcc-14.2.1-6.fc42.x86_64
libgomp-14.2.1-6.fc42.x86_64
libidn2-2.3.7-2.fc41.x86_64
libmount-2.40.2-8.fc42.x86_64
libnghttp2-1.64.0-1.fc42.x86_64
libnsl2-2.0.1-2.fc41.x86_64
libpkgconf-2.3.0-1.fc42.x86_64
libpsl-0.21.5-4.fc41.x86_64
libpwquality-1.4.5-11.fc41.x86_64
libselinux-3.7-6.fc42.x86_64
libsemanage-3.7-3.fc42.x86_64
libsepol-3.7-3.fc42.x86_64
libsmartcols-2.40.2-8.fc42.x86_64
libssh-0.11.1-1.fc42.x86_64
libssh-config-0.11.1-1.fc42.noarch
libstdc++-14.2.1-6.fc42.x86_64
libtasn1-4.19.0-9.fc41.x86_64
libtirpc-1.3.6-1.fc42.x86_64
libtool-ltdl-2.4.7-12.fc41.x86_64
libunistring-1.1-8.fc41.x86_64
libuuid-2.40.2-8.fc42.x86_64
libverto-0.3.2-9.fc41.x86_64
libxcrypt-4.4.36-10.fc42.x86_64
libxml2-2.12.8-2.fc41.x86_64
libzstd-1.5.6-2.fc41.x86_64
lua-libs-5.4.7-1.fc42.x86_64
lua-srpm-macros-1-14.fc41.noarch
lz4-libs-1.10.0-1.fc41.x86_64
mpfr-4.2.1-5.fc41.x86_64
ncurses-base-6.5-2.20240629.fc41.noarch
ncurses-libs-6.5-2.20240629.fc41.x86_64
ocaml-srpm-macros-10-3.fc41.noarch
openblas-srpm-macros-2-18.fc41.noarch
openldap-2.6.8-5.fc41.x86_64
openssl-libs-3.2.2-8.fc42.x86_64
p11-kit-0.25.5-4.fc42.x86_64
p11-kit-trust-0.25.5-4.fc42.x86_64
package-notes-srpm-macros-0.5-12.fc41.noarch
pam-1.7.0-3.fc42.x86_64
pam-libs-1.7.0-3.fc42.x86_64
patch-2.7.6-25.fc41.x86_64
pcre2-10.44-1.fc41.1.x86_64
pcre2-syntax-10.44-1.fc41.1.noarch
perl-srpm-macros-1-56.fc41.noarch
pkgconf-2.3.0-1.fc42.x86_64
pkgconf-m4-2.3.0-1.fc42.noarch
pkgconf-pkg-config-2.3.0-1.fc42.x86_64
popt-1.19-7.fc41.x86_64
publicsuffix-list-dafsa-20240107-4.fc41.noarch
pyproject-srpm-macros-1.16.2-1.fc42.noarch
python-srpm-macros-3.13-3.fc41.noarch
qt5-srpm-macros-5.15.15-1.fc42.noarch
qt6-srpm-macros-6.8.0-1.fc42.noarch
readline-8.2-11.fc42.x86_64
redhat-rpm-config-296-1.fc42.noarch
rpm-4.20.0-1.fc42.x86_64
rpm-build-4.20.0-1.fc42.x86_64
rpm-build-libs-4.20.0-1.fc42.x86_64
rpm-libs-4.20.0-1.fc42.x86_64
rpm-sequoia-1.7.0-2.fc41.x86_64
rust-srpm-macros-26.3-3.fc42.noarch
sed-4.9-3.fc41.x86_64
setup-2.15.0-5.fc41.noarch
shadow-utils-4.16.0-7.fc42.x86_64
sqlite-libs-3.47.0-1.fc42.x86_64
systemd-libs-257~rc2-2.fc42.x86_64
tar-1.35-4.fc41.x86_64
unzip-6.0-65.fc42.x86_64
util-linux-2.40.2-8.fc42.x86_64
util-linux-core-2.40.2-8.fc42.x86_64
which-2.21-42.fc41.x86_64
xxhash-libs-0.8.2-4.fc42.x86_64
xz-5.6.3-2.fc42.x86_64
xz-libs-5.6.3-2.fc42.x86_64
zig-srpm-macros-1-3.fc41.noarch
zip-3.0-41.fc41.x86_64
zlib-ng-compat-2.2.2-1.fc42.x86_64
zstd-1.5.6-2.fc41.x86_64
Start: buildsrpm
Start: rpmbuild -bs
Building target platforms: x86_64
Building for target x86_64
setting SOURCE_DATE_EPOCH=1732665600
Wrote: /builddir/build/SRPMS/llama-cpp-b4094-6.fc42.src.rpm
Finish: rpmbuild -bs
INFO: chroot_scan: 1 files copied to /var/lib/copr-rpmbuild/results/chroot_scan
INFO: /var/lib/mock/fedora-rawhide-x86_64-1732737731.306783/root/var/log/dnf5.log
INFO: chroot_scan: creating tarball /var/lib/copr-rpmbuild/results/chroot_scan.tar.gz
/bin/tar: Removing leading `/' from member names
Finish: buildsrpm
INFO: Done(/var/lib/copr-rpmbuild/workspace/workdir-a9lx8hdc/llama-cpp/llama-cpp.spec) Config(child) 0 minutes 30 seconds
INFO: Results and/or logs in: /var/lib/copr-rpmbuild/results
INFO: Cleaning up build root ('cleanup_on_success=True')
Start: clean chroot
INFO: unmounting tmpfs.
Finish: clean chroot
INFO: Start(/var/lib/copr-rpmbuild/results/llama-cpp-b4094-6.fc42.src.rpm)  Config(fedora-rawhide-x86_64)
Start(bootstrap): chroot init
INFO: mounting tmpfs at /var/lib/mock/fedora-rawhide-x86_64-bootstrap-1732737731.306783/root.
INFO: reusing tmpfs at /var/lib/mock/fedora-rawhide-x86_64-bootstrap-1732737731.306783/root.
INFO: calling preinit hooks
INFO: enabled root cache
INFO: enabled package manager cache
Start(bootstrap): cleaning package manager metadata
Finish(bootstrap): cleaning package manager metadata
Finish(bootstrap): chroot init
Start: chroot init
INFO: mounting tmpfs at /var/lib/mock/fedora-rawhide-x86_64-1732737731.306783/root.
INFO: calling preinit hooks
INFO: enabled root cache
Start: unpacking root cache
Finish: unpacking root cache
INFO: enabled package manager cache
Start: cleaning package manager metadata
Finish: cleaning package manager metadata
INFO: enabled HW Info plugin
INFO: Buildroot is handled by package management downloaded with a bootstrap image:
  rpm-4.20.0-1.fc42.x86_64
  rpm-sequoia-1.7.0-2.fc41.x86_64
  dnf5-5.2.7.0-1.fc42.x86_64
  dnf5-plugins-5.2.7.0-1.fc42.x86_64
Finish: chroot init
Start: build phase for llama-cpp-b4094-6.fc42.src.rpm
Start: build setup for llama-cpp-b4094-6.fc42.src.rpm
Building target platforms: x86_64
Building for target x86_64
setting SOURCE_DATE_EPOCH=1732665600
Wrote: /builddir/build/SRPMS/llama-cpp-b4094-6.fc42.src.rpm
Updating and loading repositories:
 fedora                                 100% | 991.9 KiB/s |  23.8 KiB |  00m00s
 Copr repository                        100% |  50.8 KiB/s |   1.5 KiB |  00m00s
Repositories loaded.
Package "curl-8.10.1-2.fc42.x86_64" is already installed.

Package                         Arch   Version                        Repository      Size
Installing:
 cmake                          x86_64 3.30.5-1.fc42                  fedora      32.7 MiB
 gcc-c++                        x86_64 14.2.1-6.fc42                  fedora      38.1 MiB
 git                            x86_64 2.47.1-1.fc42                  fedora      85.2 KiB
 hipblas-devel                  x86_64 6.2.0-8.fc42                   copr_base    2.7 MiB
 langpacks-en                   noarch 4.2-2.fc42                     fedora     400.0   B
 libcurl-devel                  x86_64 8.10.1-2.fc42                  fedora       1.3 MiB
 openmpi                        x86_64 5.0.6-1.fc42                   fedora       7.0 MiB
 pthreadpool-devel              x86_64 0.0^git20230829.4fe0e1e-5.fc41 fedora      99.1 KiB
 rocblas-devel                  x86_64 6.2.4-4.fc42                   copr_base    2.4 MiB
 rocm-comgr-devel               x86_64 18-25.rocm6.2.4.fc42           copr_base  103.1 KiB
 rocm-hip-devel                 x86_64 6.2.1-11.fc42                  copr_base    2.6 MiB
 rocm-rpm-macros                x86_64 6.2.2-1.fc42                   fedora      19.1 KiB
 rocm-rpm-macros-modules        x86_64 6.2.2-1.fc42                   fedora      24.3 KiB
 rocm-runtime-devel             x86_64 6.2.1-13.fc42                  copr_base  556.3 KiB
 wget2-wget                     x86_64 2.1.0-12.fc41                  fedora      42.0   B
 xxd                            x86_64 2:9.1.866-1.fc42               fedora      46.4 KiB
Installing dependencies:
 abattis-cantarell-vf-fonts     noarch 0.301-13.fc41                  fedora     192.7 KiB
 annobin-docs                   noarch 12.77-1.fc42                   fedora      98.4 KiB
 annobin-plugin-gcc             x86_64 12.77-1.fc42                   fedora     991.3 KiB
 brotli                         x86_64 1.1.0-5.fc41                   fedora      31.8 KiB
 brotli-devel                   x86_64 1.1.0-5.fc41                   fedora      65.6 KiB
 clang18                        x86_64 18.1.8-5.fc42                  fedora     644.4 KiB
 clang18-devel                  x86_64 18.1.8-5.fc42                  fedora      23.7 MiB
 clang18-libs                   x86_64 18.1.8-5.fc42                  fedora     102.1 MiB
 clang18-resource-filesystem    x86_64 18.1.8-5.fc42                  fedora       0.0   B
 clang18-tools-extra            x86_64 18.1.8-5.fc42                  fedora      85.3 MiB
 cmake-data                     noarch 3.30.5-1.fc42                  fedora       8.2 MiB
 cmake-filesystem               x86_64 3.30.5-1.fc42                  fedora       0.0   B
 cmake-rpm-macros               noarch 3.30.5-1.fc42                  fedora       7.5 KiB
 compiler-rt18                  x86_64 18.1.8-3.fc42                  fedora      28.0 MiB
 cpp                            x86_64 14.2.1-6.fc42                  fedora      35.0 MiB
 dbus                           x86_64 1:1.14.10-4.fc41               fedora       0.0   B
 dbus-broker                    x86_64 36-4.fc41                      fedora     382.8 KiB
 dbus-common                    noarch 1:1.14.10-4.fc41               fedora      11.2 KiB
 default-fonts-core-sans        noarch 4.2-2.fc42                     fedora      11.9 KiB
 emacs-filesystem               noarch 1:30.0-3.fc41                  fedora       0.0   B
 environment-modules            x86_64 5.5.0-1.fc42                   fedora       1.8 MiB
 expat                          x86_64 2.6.4-1.fc42                   fedora     285.5 KiB
 fonts-filesystem               noarch 1:2.0.5-17.fc41                fedora       0.0   B
 gcc                            x86_64 14.2.1-6.fc42                  fedora     104.3 MiB
 gcc-plugin-annobin             x86_64 14.2.1-6.fc42                  fedora      57.6 KiB
 git-core                       x86_64 2.47.1-1.fc42                  fedora      22.6 MiB
 git-core-doc                   noarch 2.47.1-1.fc42                  fedora      17.2 MiB
 glibc-devel                    x86_64 2.40.9000-20.fc42              fedora       2.3 MiB
 gnupg2                         x86_64 2.4.5-4.fc42                   fedora       9.6 MiB
 gnutls                         x86_64 3.8.8-1.fc42                   fedora       3.2 MiB
 gnutls-dane                    x86_64 3.8.8-1.fc42                   fedora      70.0 KiB
 google-noto-fonts-common       noarch 20240901-1.fc42                fedora      17.5 KiB
 google-noto-sans-mono-vf-fonts noarch 20240901-1.fc42                fedora     561.2 KiB
 google-noto-sans-vf-fonts      noarch 20240901-1.fc42                fedora       1.2 MiB
 google-noto-serif-vf-fonts     noarch 20240901-1.fc42                fedora       1.5 MiB
 gpgme                          x86_64 1.24.0-1.fc42                  fedora     586.7 KiB
 groff-base                     x86_64 1.23.0-7.fc41                  fedora       3.8 MiB
 hipblas                        x86_64 6.2.0-8.fc42                   copr_base  998.6 KiB
 hipcc                          x86_64 18-25.rocm6.2.4.fc42           copr_base  659.3 KiB
 hiredis                        x86_64 1.2.0-4.fc42                   fedora     114.0 KiB
 hsakmt                         x86_64 1.0.6-49.rocm6.2.4.fc42        copr_base  183.4 KiB
 hsakmt-devel                   x86_64 1.0.6-49.rocm6.2.4.fc42        copr_base  110.6 KiB
 hwdata                         noarch 0.389-1.fc42                   fedora       9.3 MiB
 hwloc-libs                     x86_64 2.11.2-1.fc42                  fedora       2.9 MiB
 jsoncpp                        x86_64 1.9.5-8.fc41                   fedora     253.4 KiB
 kernel-headers                 x86_64 6.12.0-65.fc42                 fedora       6.4 MiB
 langpacks-core-en              noarch 4.2-2.fc42                     fedora     398.0   B
 langpacks-fonts-en             noarch 4.2-2.fc42                     fedora     341.0   B
 less                           x86_64 668-1.fc42                     fedora     406.4 KiB
 libassuan                      x86_64 2.5.7-2.fc41                   fedora     163.8 KiB
 libb2                          x86_64 0.98.1-12.fc41                 fedora      42.2 KiB
 libcbor                        x86_64 0.11.0-2.fc41                  fedora      73.9 KiB
 libdrm                         x86_64 2.4.123-1.fc42                 fedora     408.0 KiB
 libedit                        x86_64 3.1-53.20240808cvs.fc41        fedora     244.1 KiB
 libedit-devel                  x86_64 3.1-53.20240808cvs.fc41        fedora      59.4 KiB
 libfabric                      x86_64 1.22.0-1.fc41                  fedora       5.2 MiB
 libfido2                       x86_64 1.15.0-2.fc41                  fedora     238.2 KiB
 libgcrypt                      x86_64 1.11.0-4.fc42                  fedora       1.5 MiB
 libgfortran                    x86_64 14.2.1-6.fc42                  fedora       3.0 MiB
 libgpg-error                   x86_64 1.50-2.fc41                    fedora     889.5 KiB
 libibverbs                     x86_64 54.0-3.fc42                    fedora       1.2 MiB
 libidn2-devel                  x86_64 2.3.7-2.fc41                   fedora     252.1 KiB
 libksba                        x86_64 1.6.7-2.fc41                   fedora     398.4 KiB
 libmpc                         x86_64 1.3.1-6.fc41                   fedora     164.7 KiB
 libnghttp2-devel               x86_64 1.64.0-1.fc42                  fedora     295.4 KiB
 libnl3                         x86_64 3.11.0-1.fc42                  fedora       1.0 MiB
 libpciaccess                   x86_64 0.16-13.fc41                   fedora      44.6 KiB
 libpipeline                    x86_64 1.5.8-1.fc42                   fedora     149.1 KiB
 libpsl-devel                   x86_64 0.21.5-4.fc41                  fedora     110.3 KiB
 libpsm2                        x86_64 12.0.1-1.fc42                  fedora     440.0 KiB
 libquadmath                    x86_64 14.2.1-6.fc42                  fedora     325.9 KiB
 librdmacm                      x86_64 54.0-3.fc42                    fedora     155.0 KiB
 libseccomp                     x86_64 2.5.5-2.fc41                   fedora     173.3 KiB
 libssh-devel                   x86_64 0.11.1-1.fc42                  fedora     177.8 KiB
 libstdc++-devel                x86_64 14.2.1-6.fc42                  fedora      15.4 MiB
 libuv                          x86_64 1:1.49.2-1.fc42                fedora     569.4 KiB
 libxcrypt-devel                x86_64 4.4.36-10.fc42                 fedora      30.5 KiB
 lld18                          x86_64 18.1.8-6.fc42                  fedora     134.5 KiB
 lld18-devel                    x86_64 18.1.8-6.fc42                  fedora      38.7 KiB
 lld18-libs                     x86_64 18.1.8-6.fc42                  fedora       5.3 MiB
 llvm18                         x86_64 18.1.8-4.fc42                  fedora     112.2 MiB
 llvm18-devel                   x86_64 18.1.8-4.fc42                  fedora      24.2 MiB
 llvm18-googletest              x86_64 18.1.8-4.fc42                  fedora       2.2 MiB
 llvm18-libs                    x86_64 18.1.8-4.fc42                  fedora     113.5 MiB
 llvm18-static                  x86_64 18.1.8-4.fc42                  fedora     283.9 MiB
 llvm18-test                    x86_64 18.1.8-4.fc42                  fedora       1.9 MiB
 logrotate                      x86_64 3.22.0-2.fc41                  fedora     153.1 KiB
 make                           x86_64 1:4.4.1-9.fc42                 fedora       1.8 MiB
 man-db                         x86_64 2.13.0-1.fc42                  fedora       2.8 MiB
 mpdecimal                      x86_64 2.5.1-16.fc41                  fedora     204.9 KiB
 munge                          x86_64 0.5.16-3.fc41                  fedora     346.3 KiB
 munge-libs                     x86_64 0.5.16-3.fc41                  fedora      32.1 KiB
 ncurses                        x86_64 6.5-2.20240629.fc41            fedora     627.3 KiB
 ncurses-c++-libs               x86_64 6.5-2.20240629.fc41            fedora     161.7 KiB
 ncurses-devel                  x86_64 6.5-2.20240629.fc41            fedora     870.1 KiB
 nettle                         x86_64 3.10-3.fc41                    fedora     793.0 KiB
 npth                           x86_64 1.8-1.fc42                     fedora      53.6 KiB
 numactl-libs                   x86_64 2.0.19-1.fc42                  fedora      54.4 KiB
 openssh                        x86_64 9.9p1-5.fc42                   fedora       1.4 MiB
 openssh-clients                x86_64 9.9p1-5.fc42                   fedora       2.7 MiB
 openssl-devel                  x86_64 1:3.2.2-8.fc42                 fedora       4.3 MiB
 orangefs                       x86_64 2.9.8-12.fc41                  fedora       3.1 MiB
 perl-AutoLoader                noarch 5.74-512.fc42                  fedora      20.5 KiB
 perl-B                         x86_64 1.89-512.fc42                  fedora     498.0 KiB
 perl-Carp                      noarch 1.54-511.fc41                  fedora      46.6 KiB
 perl-Class-Struct              noarch 0.68-512.fc42                  fedora      25.4 KiB
 perl-Data-Dumper               x86_64 2.189-512.fc41                 fedora     111.7 KiB
 perl-Digest                    noarch 1.20-511.fc41                  fedora      35.3 KiB
 perl-Digest-MD5                x86_64 2.59-5.fc41                    fedora      59.8 KiB
 perl-DynaLoader                x86_64 1.56-512.fc42                  fedora      32.1 KiB
 perl-Encode                    x86_64 4:3.21-511.fc41                fedora       4.7 MiB
 perl-Errno                     x86_64 1.38-512.fc42                  fedora       8.4 KiB
 perl-Error                     noarch 1:0.17029-16.fc41              fedora      77.3 KiB
 perl-Exporter                  noarch 5.78-511.fc41                  fedora      54.3 KiB
 perl-Fcntl                     x86_64 1.18-512.fc42                  fedora      49.0 KiB
 perl-File-Basename             noarch 2.86-512.fc42                  fedora      14.0 KiB
 perl-File-Copy                 noarch 2.41-512.fc42                  fedora      19.6 KiB
 perl-File-Find                 noarch 1.44-512.fc42                  fedora      41.9 KiB
 perl-File-Path                 noarch 2.18-511.fc41                  fedora      63.5 KiB
 perl-File-Temp                 noarch 1:0.231.100-511.fc41           fedora     162.3 KiB
 perl-File-Which                noarch 1.27-12.fc41                   fedora      30.4 KiB
 perl-File-stat                 noarch 1.14-512.fc42                  fedora      12.5 KiB
 perl-FileHandle                noarch 2.05-512.fc42                  fedora       9.3 KiB
 perl-Getopt-Long               noarch 1:2.58-2.fc41                  fedora     144.5 KiB
 perl-Getopt-Std                noarch 1.14-512.fc42                  fedora      11.2 KiB
 perl-Git                       noarch 2.47.1-1.fc42                  fedora      64.0 KiB
 perl-HTTP-Tiny                 noarch 0.090-1.fc42                   fedora     154.4 KiB
 perl-IO                        x86_64 1.55-512.fc42                  fedora     151.1 KiB
 perl-IO-Socket-IP              noarch 0.42-512.fc41                  fedora      98.7 KiB
 perl-IO-Socket-SSL             noarch 2.089-1.fc42                   fedora     703.3 KiB
 perl-IPC-Open3                 noarch 1.22-512.fc42                  fedora      22.5 KiB
 perl-MIME-Base32               noarch 1.303-21.fc41                  fedora      30.7 KiB
 perl-MIME-Base64               x86_64 3.16-511.fc41                  fedora      46.1 KiB
 perl-Net-SSLeay                x86_64 1.94-7.fc41                    fedora       1.3 MiB
 perl-POSIX                     x86_64 2.20-512.fc42                  fedora     235.1 KiB
 perl-PathTools                 x86_64 3.91-511.fc41                  fedora     180.0 KiB
 perl-Pod-Escapes               noarch 1:1.07-511.fc41                fedora      24.9 KiB
 perl-Pod-Perldoc               noarch 3.28.01-512.fc41               fedora     163.7 KiB
 perl-Pod-Simple                noarch 1:3.45-511.fc41                fedora     560.9 KiB
 perl-Pod-Usage                 noarch 4:2.03-511.fc41                fedora      84.8 KiB
 perl-Scalar-List-Utils         x86_64 5:1.68-1.fc42                  fedora     148.9 KiB
 perl-SelectSaver               noarch 1.02-512.fc42                  fedora       2.2 KiB
 perl-Socket                    x86_64 4:2.038-511.fc41               fedora     124.0 KiB
 perl-Storable                  x86_64 1:3.32-511.fc41                fedora     232.4 KiB
 perl-Symbol                    noarch 1.09-512.fc42                  fedora       6.8 KiB
 perl-Term-ANSIColor            noarch 5.01-512.fc41                  fedora      97.5 KiB
 perl-Term-Cap                  noarch 1.18-511.fc41                  fedora      29.3 KiB
 perl-TermReadKey               x86_64 2.38-23.fc41                   fedora      64.1 KiB
 perl-Text-ParseWords           noarch 3.31-511.fc41                  fedora      13.6 KiB
 perl-Text-Tabs+Wrap            noarch 2024.001-511.fc41              fedora      22.6 KiB
 perl-Time-Local                noarch 2:1.350-511.fc41               fedora      69.0 KiB
 perl-URI                       noarch 5.31-1.fc42                    fedora     257.0 KiB
 perl-base                      noarch 2.27-512.fc42                  fedora      12.5 KiB
 perl-constant                  noarch 1.33-512.fc41                  fedora      26.2 KiB
 perl-if                        noarch 0.61.000-512.fc42              fedora       5.8 KiB
 perl-interpreter               x86_64 4:5.40.0-512.fc42              fedora     122.3 KiB
 perl-lib                       x86_64 0.65-512.fc42                  fedora       8.5 KiB
 perl-libnet                    noarch 3.15-512.fc41                  fedora     289.4 KiB
 perl-libs                      x86_64 4:5.40.0-512.fc42              fedora       9.9 MiB
 perl-locale                    noarch 1.12-512.fc42                  fedora       6.5 KiB
 perl-mro                       x86_64 1.29-512.fc42                  fedora      45.6 KiB
 perl-overload                  noarch 1.37-512.fc42                  fedora      71.5 KiB
 perl-overloading               noarch 0.02-512.fc42                  fedora       4.8 KiB
 perl-parent                    noarch 1:0.242-1.fc42                 fedora      10.0 KiB
 perl-podlators                 noarch 1:6.0.2-2.fc41                 fedora     317.5 KiB
 perl-vars                      noarch 1.05-512.fc42                  fedora       3.9 KiB
 pmix                           x86_64 4.2.8-3.fc41                   fedora       2.0 MiB
 procps-ng                      x86_64 4.0.4-4.fc41                   fedora       1.0 MiB
 protobuf-c                     x86_64 1.5.0-4.fc41                   fedora      54.0 KiB
 prrte                          x86_64 3.0.6-1.fc42                   fedora     174.9 KiB
 prrte-libs                     x86_64 3.0.6-1.fc42                   fedora       1.7 MiB
 pthreadpool                    x86_64 0.0^git20230829.4fe0e1e-5.fc41 fedora     113.3 KiB
 publicsuffix-list              noarch 20240107-4.fc41                fedora     318.0 KiB
 python-pip-wheel               noarch 24.3.1-1.fc42                  fedora       1.2 MiB
 python3                        x86_64 3.13.0-1.fc42                  fedora      31.8 KiB
 python3-libs                   x86_64 3.13.0-1.fc42                  fedora      40.4 MiB
 rhash                          x86_64 1.4.5-1.fc42                   fedora     359.3 KiB
 rocblas                        x86_64 6.2.4-4.fc42                   copr_base    3.7 GiB
 rocm-comgr                     x86_64 18-25.rocm6.2.4.fc42           copr_base    8.9 MiB
 rocm-device-libs               x86_64 18-25.rocm6.2.4.fc42           copr_base    3.2 MiB
 rocm-hip                       x86_64 6.2.1-11.fc42                  copr_base   23.0 MiB
 rocm-runtime                   x86_64 6.2.1-13.fc42                  copr_base    2.7 MiB
 rocsolver                      x86_64 6.2.4-4.fc42                   copr_base  445.6 MiB
 rocsparse                      x86_64 6.2.1-3.fc42                   copr_base  860.0 MiB
 systemd                        x86_64 257~rc2-2.fc42                 fedora      17.7 MiB
 systemd-pam                    x86_64 257~rc2-2.fc42                 fedora       1.1 MiB
 systemd-rpm-macros             noarch 257~rc2-2.fc42                 fedora      10.7 KiB
 tcl                            x86_64 1:8.6.15-4.fc42                fedora       4.2 MiB
 tcsh                           x86_64 6.24.13-1.fc41                 fedora       1.2 MiB
 torque-libs                    x86_64 6.1.3-13.fc42                  fedora     458.3 KiB
 tpm2-tss                       x86_64 4.1.3-3.fc41                   fedora       1.6 MiB
 tzdata                         noarch 2024a-9.fc41                   fedora       1.7 MiB
 ucx                            x86_64 1.17.0-3.fc42                  fedora       2.4 MiB
 unbound-libs                   x86_64 1.22.0-8.fc42                  fedora       1.4 MiB
 vim-filesystem                 noarch 2:9.1.866-1.fc42               fedora      40.0   B
 wget2                          x86_64 2.1.0-12.fc41                  fedora     884.1 KiB
 wget2-libs                     x86_64 2.1.0-12.fc41                  fedora     365.7 KiB
 zlib-ng-compat-devel           x86_64 2.2.2-1.fc42                   fedora     106.8 KiB

Transaction Summary:
 Installing:       214 packages

Total size of inbound packages is 2 GiB. Need to download 2 GiB.
After this operation, 6 GiB extra will be used (install 6 GiB, remove 0 B).
[  1/214] langpacks-en-0:4.2-2.fc42.noa 100% |  54.5 KiB/s |  10.9 KiB |  00m00s
[  2/214] git-0:2.47.1-1.fc42.x86_64    100% | 226.6 KiB/s |  51.7 KiB |  00m00s
[  3/214] rocm-rpm-macros-modules-0:6.2 100% | 236.0 KiB/s |  21.0 KiB |  00m00s
[  4/214] rocm-rpm-macros-0:6.2.2-1.fc4 100% | 141.9 KiB/s |  17.2 KiB |  00m00s
[  5/214] wget2-wget-0:2.1.0-12.fc41.x8 100% |  50.5 KiB/s |   9.2 KiB |  00m00s
[  6/214] openmpi-0:5.0.6-1.fc42.x86_64 100% |   3.8 MiB/s |   2.0 MiB |  00m01s
[  7/214] xxd-2:9.1.866-1.fc42.x86_64   100% | 158.2 KiB/s |  35.9 KiB |  00m00s
[  8/214] hipblas-devel-0:6.2.0-8.fc42. 100% |   2.1 MiB/s |  88.7 KiB |  00m00s
[  9/214] libcurl-devel-0:8.10.1-2.fc42 100% |   3.1 MiB/s | 881.5 KiB |  00m00s
[ 10/214] gcc-c++-0:14.2.1-6.fc42.x86_6 100% |  35.2 MiB/s |  14.2 MiB |  00m00s
[ 11/214] pthreadpool-devel-0:0.0^git20 100% | 192.8 KiB/s |  14.3 KiB |  00m00s
[ 12/214] rocblas-devel-0:6.2.4-4.fc42. 100% |   5.2 MiB/s |  95.4 KiB |  00m00s
[ 13/214] rocm-comgr-devel-0:18-25.rocm 100% |   3.0 MiB/s |  31.2 KiB |  00m00s
[ 14/214] rocm-hip-devel-0:6.2.1-11.fc4 100% |   9.1 MiB/s | 224.1 KiB |  00m00s
[ 15/214] rocm-runtime-devel-0:6.2.1-13 100% |   5.1 MiB/s |  93.5 KiB |  00m00s
[ 16/214] cmake-0:3.30.5-1.fc42.x86_64  100% |  21.9 MiB/s |  10.9 MiB |  00m00s
[ 17/214] perl-File-Basename-0:2.86-512 100% | 237.9 KiB/s |  17.1 KiB |  00m00s
[ 18/214] perl-File-Find-0:1.44-512.fc4 100% | 616.4 KiB/s |  25.3 KiB |  00m00s
[ 19/214] git-core-doc-0:2.47.1-1.fc42. 100% |  20.2 MiB/s |   3.0 MiB |  00m00s
[ 20/214] perl-Getopt-Long-1:2.58-2.fc4 100% |   1.3 MiB/s |  63.9 KiB |  00m00s
[ 21/214] perl-Git-0:2.47.1-1.fc42.noar 100% | 817.8 KiB/s |  38.4 KiB |  00m00s
[ 22/214] git-core-0:2.47.1-1.fc42.x86_ 100% |  23.2 MiB/s |   4.9 MiB |  00m00s
[ 23/214] perl-IPC-Open3-0:1.22-512.fc4 100% | 778.9 KiB/s |  21.8 KiB |  00m00s
[ 24/214] perl-PathTools-0:3.91-511.fc4 100% |   2.9 MiB/s |  87.4 KiB |  00m00s
[ 25/214] perl-TermReadKey-0:2.38-23.fc 100% | 935.5 KiB/s |  35.6 KiB |  00m00s
[ 26/214] perl-interpreter-4:5.40.0-512 100% |   1.9 MiB/s |  72.3 KiB |  00m00s
[ 27/214] perl-lib-0:0.65-512.fc42.x86_ 100% | 413.9 KiB/s |  14.9 KiB |  00m00s
[ 28/214] langpacks-core-en-0:4.2-2.fc4 100% | 279.7 KiB/s |  10.9 KiB |  00m00s
[ 29/214] langpacks-fonts-en-0:4.2-2.fc 100% | 243.9 KiB/s |  11.2 KiB |  00m00s
[ 30/214] hwloc-libs-0:2.11.2-1.fc42.x8 100% |  31.2 MiB/s |   2.1 MiB |  00m00s
[ 31/214] libfabric-0:1.22.0-1.fc41.x86 100% |  20.6 MiB/s |   1.4 MiB |  00m00s
[ 32/214] libgfortran-0:14.2.1-6.fc42.x 100% |  13.7 MiB/s | 939.2 KiB |  00m00s
[ 33/214] libpsm2-0:12.0.1-1.fc42.x86_6 100% |   4.8 MiB/s | 200.0 KiB |  00m00s
[ 34/214] libquadmath-0:14.2.1-6.fc42.x 100% |   2.7 MiB/s | 204.3 KiB |  00m00s
[ 35/214] openssh-clients-0:9.9p1-5.fc4 100% |   8.5 MiB/s | 761.6 KiB |  00m00s
[ 36/214] orangefs-0:2.9.8-12.fc41.x86_ 100% |  18.3 MiB/s |   1.8 MiB |  00m00s
[ 37/214] pmix-0:4.2.8-3.fc41.x86_64    100% |  10.9 MiB/s | 668.9 KiB |  00m00s
[ 38/214] prrte-0:3.0.6-1.fc42.x86_64   100% |   1.3 MiB/s |  56.8 KiB |  00m00s
[ 39/214] ucx-0:1.17.0-3.fc42.x86_64    100% |  12.7 MiB/s | 835.4 KiB |  00m00s
[ 40/214] environment-modules-0:5.5.0-1 100% |  10.2 MiB/s | 764.6 KiB |  00m00s
[ 41/214] wget2-0:2.1.0-12.fc41.x86_64  100% |   3.4 MiB/s | 249.0 KiB |  00m00s
[ 42/214] cmake-data-0:3.30.5-1.fc42.no 100% |  37.8 MiB/s |   2.3 MiB |  00m00s
[ 43/214] cmake-filesystem-0:3.30.5-1.f 100% | 458.9 KiB/s |  17.4 KiB |  00m00s
[ 44/214] expat-0:2.6.4-1.fc42.x86_64   100% |   2.9 MiB/s | 114.5 KiB |  00m00s
[ 45/214] jsoncpp-0:1.9.5-8.fc41.x86_64 100% |   1.8 MiB/s |  99.3 KiB |  00m00s
[ 46/214] libuv-1:1.49.2-1.fc42.x86_64  100% |   3.8 MiB/s | 263.7 KiB |  00m00s
[ 47/214] make-1:4.4.1-9.fc42.x86_64    100% |   8.3 MiB/s | 586.3 KiB |  00m00s
[ 48/214] rhash-0:1.4.5-1.fc42.x86_64   100% |   4.4 MiB/s | 198.0 KiB |  00m00s
[ 49/214] hipblas-0:6.2.0-8.fc42.x86_64 100% |   8.6 MiB/s | 159.1 KiB |  00m00s
[ 50/214] libmpc-0:1.3.1-6.fc41.x86_64  100% | 748.6 KiB/s |  71.1 KiB |  00m00s
[ 51/214] pthreadpool-0:0.0^git20230829 100% | 708.7 KiB/s |  46.1 KiB |  00m00s
[ 52/214] perl-File-Copy-0:2.41-512.fc4 100% | 180.8 KiB/s |  20.1 KiB |  00m00s
[ 53/214] perl-File-Which-0:1.27-12.fc4 100% | 417.2 KiB/s |  21.7 KiB |  00m00s
[ 54/214] perl-Getopt-Std-0:1.14-512.fc 100% | 474.2 KiB/s |  15.6 KiB |  00m00s
[ 55/214] perl-Scalar-List-Utils-5:1.68 100% | 431.3 KiB/s |  74.2 KiB |  00m00s
[ 56/214] perl-URI-0:5.31-1.fc42.noarch 100% | 291.8 KiB/s | 140.6 KiB |  00m00s
[ 57/214] rocm-hip-0:6.2.1-11.fc42.x86_ 100% |  19.0 MiB/s |   9.3 MiB |  00m00s
[ 58/214] rocm-runtime-0:6.2.1-13.fc42. 100% |   2.9 MiB/s | 545.6 KiB |  00m00s
[ 59/214] less-0:668-1.fc42.x86_64      100% | 607.0 KiB/s | 189.4 KiB |  00m00s
[ 60/214] perl-Carp-0:1.54-511.fc41.noa 100% | 140.2 KiB/s |  28.9 KiB |  00m00s
[ 61/214] gcc-0:14.2.1-6.fc42.x86_64    100% |  15.8 MiB/s |  37.0 MiB |  00m02s
[ 62/214] perl-Exporter-0:5.78-511.fc41 100% | 161.0 KiB/s |  30.9 KiB |  00m00s
[ 63/214] perl-Pod-Usage-4:2.03-511.fc4 100% |   1.2 MiB/s |  40.0 KiB |  00m00s
[ 64/214] perl-Text-ParseWords-0:3.31-5 100% | 376.7 KiB/s |  16.6 KiB |  00m00s
[ 65/214] perl-base-0:2.27-512.fc42.noa 100% | 448.8 KiB/s |  16.2 KiB |  00m00s
[ 66/214] perl-overload-0:1.37-512.fc42 100% |   1.2 MiB/s |  45.5 KiB |  00m00s
[ 67/214] perl-Error-1:0.17029-16.fc41. 100% | 386.7 KiB/s |  40.6 KiB |  00m00s
[ 68/214] perl-Fcntl-0:1.18-512.fc42.x8 100% | 783.4 KiB/s |  29.8 KiB |  00m00s
[ 69/214] perl-IO-0:1.55-512.fc42.x86_6 100% |   2.1 MiB/s |  81.7 KiB |  00m00s
[ 70/214] perl-POSIX-0:2.20-512.fc42.x8 100% |   2.3 MiB/s |  97.0 KiB |  00m00s
[ 71/214] perl-constant-0:1.33-512.fc41 100% |  62.6 KiB/s |  23.0 KiB |  00m00s
[ 72/214] perl-Errno-0:1.38-512.fc42.x8 100% | 252.7 KiB/s |  14.9 KiB |  00m00s
[ 73/214] perl-Symbol-0:1.09-512.fc42.n 100% |  36.8 KiB/s |  14.2 KiB |  00m00s
[ 74/214] perl-DynaLoader-0:1.56-512.fc 100% |  36.2 KiB/s |  26.0 KiB |  00m01s
[ 75/214] perl-vars-0:1.05-512.fc42.noa 100% |  86.4 KiB/s |  13.0 KiB |  00m00s
[ 76/214] default-fonts-core-sans-0:4.2 100% | 129.7 KiB/s |  31.3 KiB |  00m00s
[ 77/214] perl-libs-4:5.40.0-512.fc42.x 100% |   1.3 MiB/s |   2.3 MiB |  00m02s
[ 78/214] google-noto-sans-mono-vf-font 100% | 492.4 KiB/s | 278.2 KiB |  00m01s
[ 79/214] google-noto-serif-vf-fonts-0: 100% |   1.9 MiB/s | 645.8 KiB |  00m00s
[ 80/214] libnl3-0:3.11.0-1.fc42.x86_64 100% |   2.0 MiB/s | 353.1 KiB |  00m00s
[ 81/214] librdmacm-0:54.0-3.fc42.x86_6 100% | 955.9 KiB/s |  70.7 KiB |  00m00s
[ 82/214] libibverbs-0:54.0-3.fc42.x86_ 100% | 852.9 KiB/s | 440.1 KiB |  00m01s
[ 83/214] libedit-0:3.1-53.20240808cvs. 100% | 771.0 KiB/s | 105.6 KiB |  00m00s
[ 84/214] numactl-libs-0:2.0.19-1.fc42. 100% |  93.5 KiB/s |  31.1 KiB |  00m00s
[ 85/214] libfido2-0:1.15.0-2.fc41.x86_ 100% | 317.6 KiB/s |  98.1 KiB |  00m00s
[ 86/214] openssh-0:9.9p1-5.fc42.x86_64 100% | 874.5 KiB/s | 353.3 KiB |  00m00s
[ 87/214] munge-libs-0:0.5.16-3.fc41.x8 100% | 375.4 KiB/s |  21.4 KiB |  00m00s
[ 88/214] tcsh-0:6.24.13-1.fc41.x86_64  100% |   1.2 MiB/s | 459.5 KiB |  00m00s
[ 89/214] prrte-libs-0:3.0.6-1.fc42.x86 100% |   1.3 MiB/s | 550.0 KiB |  00m00s
[ 90/214] vim-filesystem-2:9.1.866-1.fc 100% | 155.1 KiB/s |  16.4 KiB |  00m00s
[ 91/214] gpgme-0:1.24.0-1.fc42.x86_64  100% |   1.7 MiB/s | 217.7 KiB |  00m00s
[ 92/214] rocblas-0:6.2.4-4.fc42.x86_64 100% |  73.8 MiB/s | 538.5 MiB |  00m07s
[ 93/214] wget2-libs-0:2.1.0-12.fc41.x8 100% | 200.7 KiB/s | 146.3 KiB |  00m01s
[ 94/214] man-db-0:2.13.0-1.fc42.x86_64 100% |   1.0 MiB/s |   1.3 MiB |  00m01s
[ 95/214] emacs-filesystem-1:30.0-3.fc4 100% | 140.0 KiB/s |   7.1 KiB |  00m00s
[ 96/214] perl-Data-Dumper-0:2.189-512. 100% | 761.4 KiB/s |  56.3 KiB |  00m00s
[ 97/214] perl-MIME-Base32-0:1.303-21.f 100% | 487.4 KiB/s |  20.5 KiB |  00m00s
[ 98/214] perl-libnet-0:3.15-512.fc41.n 100% |   1.7 MiB/s | 128.5 KiB |  00m00s
[ 99/214] perl-MIME-Base64-0:3.16-511.f 100% | 230.3 KiB/s |  29.9 KiB |  00m00s
[100/214] perl-parent-1:0.242-1.fc42.no 100% | 404.3 KiB/s |  15.0 KiB |  00m00s
[101/214] perl-Pod-Perldoc-0:3.28.01-51 100% |   1.2 MiB/s |  86.1 KiB |  00m00s
[102/214] libdrm-0:2.4.123-1.fc42.x86_6 100% |   1.3 MiB/s | 158.5 KiB |  00m00s
[103/214] perl-podlators-1:6.0.2-2.fc41 100% |   1.6 MiB/s | 128.8 KiB |  00m00s
[104/214] perl-mro-0:1.29-512.fc42.x86_ 100% | 586.0 KiB/s |  29.9 KiB |  00m00s
[105/214] perl-overloading-0:0.02-512.f 100% | 313.6 KiB/s |  12.9 KiB |  00m00s
[106/214] perl-SelectSaver-0:1.02-512.f 100% | 315.6 KiB/s |  11.7 KiB |  00m00s
[107/214] perl-File-stat-0:1.14-512.fc4 100% | 293.2 KiB/s |  17.0 KiB |  00m00s
[108/214] perl-locale-0:1.12-512.fc42.n 100% | 399.2 KiB/s |  13.6 KiB |  00m00s
[109/214] perl-Socket-4:2.038-511.fc41. 100% | 856.7 KiB/s |  54.8 KiB |  00m00s
[110/214] abattis-cantarell-vf-fonts-0: 100% |   1.4 MiB/s | 120.2 KiB |  00m00s
[111/214] cpp-0:14.2.1-6.fc42.x86_64    100% |  19.4 MiB/s |  11.9 MiB |  00m01s
[112/214] fonts-filesystem-1:2.0.5-17.f 100% | 146.3 KiB/s |   8.5 KiB |  00m00s
[113/214] google-noto-fonts-common-0:20 100% | 413.3 KiB/s |  17.8 KiB |  00m00s
[114/214] libcbor-0:0.11.0-2.fc41.x86_6 100% | 662.7 KiB/s |  33.1 KiB |  00m00s
[115/214] groff-base-0:1.23.0-7.fc41.x8 100% |  12.9 MiB/s |   1.1 MiB |  00m00s
[116/214] torque-libs-0:6.1.3-13.fc42.x 100% |   1.6 MiB/s | 187.9 KiB |  00m00s
[117/214] libpipeline-0:1.5.8-1.fc42.x8 100% |   1.4 MiB/s |  59.7 KiB |  00m00s
[118/214] google-noto-sans-vf-fonts-0:2 100% |   1.9 MiB/s | 593.9 KiB |  00m00s
[119/214] libassuan-0:2.5.7-2.fc41.x86_ 100% |   1.5 MiB/s |  67.1 KiB |  00m00s
[120/214] gnutls-0:3.8.8-1.fc42.x86_64  100% |  16.7 MiB/s |   1.1 MiB |  00m00s
[121/214] libgpg-error-0:1.50-2.fc41.x8 100% |   1.9 MiB/s | 237.5 KiB |  00m00s
[122/214] gnutls-dane-0:3.8.8-1.fc42.x8 100% |   1.3 MiB/s |  43.5 KiB |  00m00s
[123/214] perl-Digest-MD5-0:2.59-5.fc41 100% |   1.2 MiB/s |  36.0 KiB |  00m00s
[124/214] perl-FileHandle-0:2.05-512.fc 100% | 532.8 KiB/s |  15.5 KiB |  00m00s
[125/214] perl-B-0:1.89-512.fc42.x86_64 100% |   2.2 MiB/s | 176.3 KiB |  00m00s
[126/214] perl-IO-Socket-IP-0:0.42-512. 100% |   1.4 MiB/s |  41.8 KiB |  00m00s
[127/214] perl-Time-Local-2:1.350-511.f 100% | 933.1 KiB/s |  34.5 KiB |  00m00s
[128/214] libpciaccess-0:0.16-13.fc41.x 100% | 678.9 KiB/s |  26.5 KiB |  00m00s
[129/214] perl-File-Temp-1:0.231.100-51 100% |   1.3 MiB/s |  59.1 KiB |  00m00s
[130/214] perl-HTTP-Tiny-0:0.090-1.fc42 100% |   1.9 MiB/s |  56.5 KiB |  00m00s
[131/214] perl-Term-ANSIColor-0:5.01-51 100% |   1.7 MiB/s |  47.7 KiB |  00m00s
[132/214] perl-Term-Cap-0:1.18-511.fc41 100% | 788.2 KiB/s |  22.1 KiB |  00m00s
[133/214] perl-Pod-Simple-1:3.45-511.fc 100% |   2.4 MiB/s | 219.0 KiB |  00m00s
[134/214] perl-Class-Struct-0:0.68-512. 100% | 786.2 KiB/s |  22.0 KiB |  00m00s
[135/214] munge-0:0.5.16-3.fc41.x86_64  100% |   1.8 MiB/s | 133.8 KiB |  00m00s
[136/214] libgcrypt-0:1.11.0-4.fc42.x86 100% |   6.5 MiB/s | 583.2 KiB |  00m00s
[137/214] gnupg2-0:2.4.5-4.fc42.x86_64  100% |   5.3 MiB/s |   2.7 MiB |  00m01s
[138/214] libksba-0:1.6.7-2.fc41.x86_64 100% |   1.5 MiB/s | 159.7 KiB |  00m00s
[139/214] npth-0:1.8-1.fc42.x86_64      100% | 275.6 KiB/s |  25.9 KiB |  00m00s
[140/214] tpm2-tss-0:4.1.3-3.fc41.x86_6 100% |   4.2 MiB/s | 411.5 KiB |  00m00s
[141/214] perl-if-0:0.61.000-512.fc42.n 100% | 214.8 KiB/s |  14.0 KiB |  00m00s
[142/214] unbound-libs-0:1.22.0-8.fc42. 100% |   6.5 MiB/s | 554.2 KiB |  00m00s
[143/214] perl-Digest-0:1.20-511.fc41.n 100% | 830.1 KiB/s |  24.9 KiB |  00m00s
[144/214] perl-File-Path-0:2.18-511.fc4 100% | 750.2 KiB/s |  35.3 KiB |  00m00s
[145/214] nettle-0:3.10-3.fc41.x86_64   100% |   2.1 MiB/s | 428.5 KiB |  00m00s
[146/214] hwdata-0:0.389-1.fc42.noarch  100% |  15.4 MiB/s |   1.6 MiB |  00m00s
[147/214] perl-IO-Socket-SSL-0:2.089-1. 100% |   4.9 MiB/s | 231.2 KiB |  00m00s
[148/214] perl-Pod-Escapes-1:1.07-511.f 100% | 707.6 KiB/s |  19.8 KiB |  00m00s
[149/214] perl-Text-Tabs+Wrap-0:2024.00 100% | 780.4 KiB/s |  21.9 KiB |  00m00s
[150/214] ncurses-0:6.5-2.20240629.fc41 100% |   8.1 MiB/s | 423.8 KiB |  00m00s
[151/214] logrotate-0:3.22.0-2.fc41.x86 100% |   1.5 MiB/s |  75.9 KiB |  00m00s
[152/214] perl-Net-SSLeay-0:1.94-7.fc41 100% |   3.6 MiB/s | 375.7 KiB |  00m00s
[153/214] hiredis-0:1.2.0-4.fc42.x86_64 100% |   1.7 MiB/s |  49.9 KiB |  00m00s
[154/214] protobuf-c-0:1.5.0-4.fc41.x86 100% | 809.9 KiB/s |  32.4 KiB |  00m00s
[155/214] perl-AutoLoader-0:5.74-512.fc 100% | 756.5 KiB/s |  21.2 KiB |  00m00s
[156/214] libb2-0:0.98.1-12.fc41.x86_64 100% | 885.4 KiB/s |  25.7 KiB |  00m00s
[157/214] mpdecimal-0:2.5.1-16.fc41.x86 100% |   2.9 MiB/s |  89.0 KiB |  00m00s
[158/214] tzdata-0:2024a-9.fc41.noarch  100% |  14.0 MiB/s | 714.7 KiB |  00m00s
[159/214] hsakmt-devel-0:1.0.6-49.rocm6 100% |   3.3 MiB/s |  37.4 KiB |  00m00s
[160/214] hsakmt-0:1.0.6-49.rocm6.2.4.f 100% |   3.8 MiB/s |  73.7 KiB |  00m00s
[161/214] hipcc-0:18-25.rocm6.2.4.fc42. 100% |   6.9 MiB/s | 141.9 KiB |  00m00s
[162/214] python-pip-wheel-0:24.3.1-1.f 100% |   9.8 MiB/s |   1.2 MiB |  00m00s
[163/214] clang18-resource-filesystem-0 100% | 241.6 KiB/s |  13.5 KiB |  00m00s
[164/214] rocm-comgr-0:18-25.rocm6.2.4. 100% |  39.5 MiB/s |   2.8 MiB |  00m00s
[165/214] compiler-rt18-0:18.1.8-3.fc42 100% |  13.8 MiB/s |   2.3 MiB |  00m00s
[166/214] lld18-libs-0:18.1.8-6.fc42.x8 100% |  16.1 MiB/s |   1.5 MiB |  00m00s
[167/214] python3-libs-0:3.13.0-1.fc42. 100% |   8.7 MiB/s |   9.1 MiB |  00m01s
[168/214] llvm18-libs-0:18.1.8-4.fc42.x 100% |  31.3 MiB/s |  28.0 MiB |  00m01s
[169/214] clang18-devel-0:18.1.8-5.fc42 100% |   9.8 MiB/s |   3.2 MiB |  00m00s
[170/214] clang18-0:18.1.8-5.fc42.x86_6 100% |   1.2 MiB/s |  72.8 KiB |  00m00s
[171/214] llvm18-devel-0:18.1.8-4.fc42. 100% |  11.2 MiB/s |   4.0 MiB |  00m00s
[172/214] clang18-libs-0:18.1.8-5.fc42. 100% |  11.6 MiB/s |  21.8 MiB |  00m02s
[173/214] llvm18-googletest-0:18.1.8-4. 100% |   3.4 MiB/s | 391.0 KiB |  00m00s
[174/214] clang18-tools-extra-0:18.1.8- 100% |  13.4 MiB/s |  19.7 MiB |  00m01s
[175/214] llvm18-test-0:18.1.8-4.fc42.x 100% |   6.0 MiB/s | 651.0 KiB |  00m00s
[176/214] rocm-device-libs-0:18-25.rocm 100% |  22.8 MiB/s | 560.1 KiB |  00m00s
[177/214] llvm18-0:18.1.8-4.fc42.x86_64 100% |  14.7 MiB/s |  26.9 MiB |  00m02s
[178/214] perl-Encode-4:3.21-511.fc41.x 100% |   4.5 MiB/s |   1.1 MiB |  00m00s
[179/214] perl-Storable-1:3.32-511.fc41 100% |   1.0 MiB/s |  98.4 KiB |  00m00s
[180/214] systemd-0:257~rc2-2.fc42.x86_ 100% |  19.3 MiB/s |   5.9 MiB |  00m00s
[181/214] dbus-1:1.14.10-4.fc41.x86_64  100% | 193.0 KiB/s |   7.9 KiB |  00m00s
[182/214] llvm18-static-0:18.1.8-4.fc42 100% |  19.0 MiB/s |  38.0 MiB |  00m02s
[183/214] libseccomp-0:2.5.5-2.fc41.x86 100% |   1.2 MiB/s |  70.2 KiB |  00m00s
[184/214] systemd-pam-0:257~rc2-2.fc42. 100% |   5.2 MiB/s | 417.1 KiB |  00m00s
[185/214] dbus-common-1:1.14.10-4.fc41. 100% | 318.7 KiB/s |  14.7 KiB |  00m00s
[186/214] libedit-devel-0:3.1-53.202408 100% | 668.5 KiB/s |  40.8 KiB |  00m00s
[187/214] dbus-broker-0:36-4.fc41.x86_6 100% | 617.5 KiB/s | 171.7 KiB |  00m00s
[188/214] ncurses-devel-0:6.5-2.2024062 100% |   4.2 MiB/s | 569.9 KiB |  00m00s
[189/214] ncurses-c++-libs-0:6.5-2.2024 100% | 630.1 KiB/s |  37.8 KiB |  00m00s
[190/214] python3-0:3.13.0-1.fc42.x86_6 100% | 867.4 KiB/s |  27.8 KiB |  00m00s
[191/214] brotli-devel-0:1.1.0-5.fc41.x 100% | 459.6 KiB/s |  34.0 KiB |  00m00s
[192/214] brotli-0:1.1.0-5.fc41.x86_64  100% | 304.2 KiB/s |  20.1 KiB |  00m00s
[193/214] libidn2-devel-0:2.3.7-2.fc41. 100% | 983.0 KiB/s |  70.8 KiB |  00m00s
[194/214] libnghttp2-devel-0:1.64.0-1.f 100% | 774.1 KiB/s |  55.7 KiB |  00m00s
[195/214] libpsl-devel-0:0.21.5-4.fc41. 100% | 875.4 KiB/s |  33.3 KiB |  00m00s
[196/214] publicsuffix-list-0:20240107- 100% |   1.6 MiB/s |  87.5 KiB |  00m00s
[197/214] libssh-devel-0:0.11.1-1.fc42. 100% | 578.2 KiB/s |  42.2 KiB |  00m00s
[198/214] zlib-ng-compat-devel-0:2.2.2- 100% | 657.9 KiB/s |  38.2 KiB |  00m00s
[199/214] openssl-devel-1:3.2.2-8.fc42. 100% |  11.2 MiB/s |   2.8 MiB |  00m00s
[200/214] libstdc++-devel-0:14.2.1-6.fc 100% |  15.6 MiB/s |   2.8 MiB |  00m00s
[201/214] libxcrypt-devel-0:4.4.36-10.f 100% | 589.9 KiB/s |  28.9 KiB |  00m00s
[202/214] glibc-devel-0:2.40.9000-20.fc 100% |   5.4 MiB/s | 644.1 KiB |  00m00s
[203/214] procps-ng-0:4.0.4-4.fc41.x86_ 100% |   6.0 MiB/s | 366.8 KiB |  00m00s
[204/214] tcl-1:8.6.15-4.fc42.x86_64    100% |   8.0 MiB/s |   1.1 MiB |  00m00s
[205/214] lld18-devel-0:18.1.8-6.fc42.x 100% | 601.4 KiB/s |  24.7 KiB |  00m00s
[206/214] lld18-0:18.1.8-6.fc42.x86_64  100% | 537.5 KiB/s |  26.9 KiB |  00m00s
[207/214] kernel-headers-0:6.12.0-65.fc 100% |  10.7 MiB/s |   1.6 MiB |  00m00s
[208/214] annobin-plugin-gcc-0:12.77-1. 100% |   6.6 MiB/s | 977.2 KiB |  00m00s
[209/214] gcc-plugin-annobin-0:14.2.1-6 100% |   1.3 MiB/s |  57.1 KiB |  00m00s
[210/214] annobin-docs-0:12.77-1.fc42.n 100% |   1.7 MiB/s |  92.4 KiB |  00m00s
[211/214] systemd-rpm-macros-0:257~rc2- 100% | 795.3 KiB/s |  34.2 KiB |  00m00s
[212/214] cmake-rpm-macros-0:3.30.5-1.f 100% | 329.4 KiB/s |  16.8 KiB |  00m00s
[213/214] rocsolver-0:6.2.4-4.fc42.x86_ 100% |  56.5 MiB/s | 375.7 MiB |  00m07s
[214/214] rocsparse-0:6.2.1-3.fc42.x86_ 100% |  68.0 MiB/s | 797.5 MiB |  00m12s
--------------------------------------------------------------------------------
[214/214] Total                         100% |  73.0 MiB/s |   2.0 GiB |  00m28s
Running transaction
[  1/216] Verify package files          100% |  28.0   B/s | 214.0   B |  00m08s
[  2/216] Prepare transaction           100% | 743.0   B/s | 214.0   B |  00m00s
[  3/216] Installing cmake-filesystem-0 100% |   2.4 MiB/s |   7.3 KiB |  00m00s
[  4/216] Installing libgpg-error-0:1.5 100% | 124.9 MiB/s | 895.4 KiB |  00m00s
[  5/216] Installing hwloc-libs-0:2.11. 100% | 286.6 MiB/s |   2.9 MiB |  00m00s
[  6/216] Installing fonts-filesystem-1 100% |   0.0   B/s | 788.0   B |  00m00s
[  7/216] Installing numactl-libs-0:2.0 100% |  54.0 MiB/s |  55.3 KiB |  00m00s
[  8/216] Installing expat-0:2.6.4-1.fc 100% | 140.4 MiB/s | 287.6 KiB |  00m00s
[  9/216] Installing clang18-resource-f 100% | 980.5 KiB/s |   1.0 KiB |  00m00s
[ 10/216] Installing google-noto-fonts- 100% |  17.8 MiB/s |  18.3 KiB |  00m00s
[ 11/216] Installing munge-libs-0:0.5.1 100% |  32.2 MiB/s |  32.9 KiB |  00m00s
[ 12/216] Installing pmix-0:4.2.8-3.fc4 100% | 201.6 MiB/s |   2.0 MiB |  00m00s
[ 13/216] Installing libedit-0:3.1-53.2 100% | 120.0 MiB/s | 245.8 KiB |  00m00s
[ 14/216] Installing llvm18-libs-0:18.1 100% | 297.9 MiB/s | 113.5 MiB |  00m00s
[ 15/216] Installing clang18-libs-0:18. 100% | 343.9 MiB/s | 102.1 MiB |  00m00s
[ 16/216] Installing lld18-libs-0:18.1. 100% | 309.4 MiB/s |   5.3 MiB |  00m00s
[ 17/216] Installing libnl3-0:3.11.0-1. 100% | 171.1 MiB/s |   1.0 MiB |  00m00s
[ 18/216] Installing libibverbs-0:54.0- 100% | 136.5 MiB/s |   1.2 MiB |  00m00s
[ 19/216] Installing less-0:668-1.fc42. 100% | 133.4 MiB/s | 409.7 KiB |  00m00s
[ 20/216] Installing libmpc-0:1.3.1-6.f 100% |  81.1 MiB/s | 166.2 KiB |  00m00s
[ 21/216] Installing rocm-comgr-0:18-25 100% | 298.0 MiB/s |   8.9 MiB |  00m00s
[ 22/216] Installing libpsm2-0:12.0.1-1 100% | 143.6 MiB/s | 441.1 KiB |  00m00s
[ 23/216] Installing libassuan-0:2.5.7- 100% |  40.4 MiB/s | 165.6 KiB |  00m00s
[ 24/216] Installing libstdc++-devel-0: 100% | 202.0 MiB/s |  15.6 MiB |  00m00s
[ 25/216] Installing nettle-0:3.10-3.fc 100% | 194.4 MiB/s | 796.1 KiB |  00m00s
[ 26/216] Installing gnutls-0:3.8.8-1.f 100% | 120.3 MiB/s |   3.2 MiB |  00m00s
[ 27/216] Installing groff-base-0:1.23. 100% | 107.6 MiB/s |   3.9 MiB |  00m00s
[ 28/216] Installing emacs-filesystem-1 100% | 531.2 KiB/s | 544.0   B |  00m00s
[ 29/216] Installing vim-filesystem-2:9 100% |   2.3 MiB/s |   4.7 KiB |  00m00s
[ 30/216] Installing make-1:4.4.1-9.fc4 100% | 163.7 MiB/s |   1.8 MiB |  00m00s
[ 31/216] Installing jsoncpp-0:1.9.5-8. 100% | 124.5 MiB/s | 254.9 KiB |  00m00s
[ 32/216] Installing cpp-0:14.2.1-6.fc4 100% | 273.3 MiB/s |  35.0 MiB |  00m00s
[ 33/216] Installing librdmacm-0:54.0-3 100% |  76.7 MiB/s | 157.0 KiB |  00m00s
[ 34/216] Installing libfabric-0:1.22.0 100% | 322.8 MiB/s |   5.2 MiB |  00m00s
[ 35/216] Installing lld18-0:18.1.8-6.f 100% |  67.1 MiB/s | 137.4 KiB |  00m00s
[ 36/216] Installing lld18-devel-0:18.1 100% |  21.0 MiB/s |  43.0 KiB |  00m00s
[ 37/216] Installing google-noto-sans-m 100% | 183.0 MiB/s | 562.2 KiB |  00m00s
[ 38/216] Installing google-noto-serif- 100% | 253.3 MiB/s |   1.5 MiB |  00m00s
[ 39/216] Installing google-noto-sans-v 100% | 208.2 MiB/s |   1.2 MiB |  00m00s
[ 40/216] Installing abattis-cantarell- 100% |  94.9 MiB/s | 194.4 KiB |  00m00s
[ 41/216] Installing default-fonts-core 100% |   8.9 MiB/s |  18.2 KiB |  00m00s
[ 42/216] Installing langpacks-core-en- 100% |   0.0   B/s | 704.0   B |  00m00s
[ 43/216] Installing langpacks-fonts-en 100% |   0.0   B/s | 652.0   B |  00m00s
[ 44/216] Installing libgcrypt-0:1.11.0 100% | 258.3 MiB/s |   1.6 MiB |  00m00s
[ 45/216] Installing libksba-0:1.6.7-2. 100% | 130.5 MiB/s | 401.0 KiB |  00m00s
[ 46/216] Installing libssh-devel-0:0.1 100% | 176.1 MiB/s | 180.3 KiB |  00m00s
[ 47/216] Installing zlib-ng-compat-dev 100% |  52.9 MiB/s | 108.3 KiB |  00m00s
[ 48/216] Installing annobin-docs-0:12. 100% |  32.4 MiB/s |  99.5 KiB |  00m00s
[ 49/216] Installing kernel-headers-0:6 100% | 108.1 MiB/s |   6.6 MiB |  00m00s
[ 50/216] Installing libxcrypt-devel-0: 100% |  10.7 MiB/s |  32.9 KiB |  00m00s
[ 51/216] Installing glibc-devel-0:2.40 100% |  80.3 MiB/s |   2.3 MiB |  00m00s
[ 52/216] Installing gcc-0:14.2.1-6.fc4 100% | 297.2 MiB/s | 104.3 MiB |  00m00s
[ 53/216] Installing gcc-c++-0:14.2.1-6 100% | 280.4 MiB/s |  38.1 MiB |  00m00s
[ 54/216] Installing clang18-0:18.1.8-5 100% | 210.6 MiB/s | 647.1 KiB |  00m00s
[ 55/216] Installing tcl-1:8.6.15-4.fc4 100% | 178.3 MiB/s |   4.3 MiB |  00m00s
[ 56/216] Installing procps-ng-0:4.0.4- 100% |  65.1 MiB/s |   1.0 MiB |  00m00s
[ 57/216] Installing openssl-devel-1:3. 100% |  29.3 MiB/s |   5.2 MiB |  00m00s
[ 58/216] Installing publicsuffix-list- 100% | 311.6 MiB/s | 319.1 KiB |  00m00s
[ 59/216] Installing libpsl-devel-0:0.2 100% |  55.5 MiB/s | 113.6 KiB |  00m00s
[ 60/216] Installing libnghttp2-devel-0 100% | 144.8 MiB/s | 296.5 KiB |  00m00s
[ 61/216] Installing libidn2-devel-0:2. 100% |  84.7 MiB/s | 260.1 KiB |  00m00s
[ 62/216] Installing brotli-0:1.1.0-5.f 100% |  31.7 MiB/s |  32.5 KiB |  00m00s
[ 63/216] Installing brotli-devel-0:1.1 100% |  66.4 MiB/s |  68.0 KiB |  00m00s
[ 64/216] Installing ncurses-c++-libs-0 100% |  79.5 MiB/s | 162.9 KiB |  00m00s
[ 65/216] Installing ncurses-devel-0:6. 100% |  31.0 MiB/s |   1.0 MiB |  00m00s
[ 66/216] Installing libedit-devel-0:3. 100% |  21.8 MiB/s |  67.0 KiB |  00m00s
[ 67/216] Installing dbus-common-1:1.14 100% | 366.2 KiB/s |  13.6 KiB |  00m00s
[ 68/216] Installing dbus-broker-0:36-4 100% |  26.9 MiB/s | 385.3 KiB |  00m00s
[ 69/216] Installing dbus-1:1.14.10-4.f 100% |   0.0   B/s | 124.0   B |  00m00s
[ 70/216] Installing libseccomp-0:2.5.5 100% |  85.5 MiB/s | 175.2 KiB |  00m00s
[ 71/216] Installing systemd-pam-0:257~ 100% | 141.1 MiB/s |   1.1 MiB |  00m00s
[ 72/216] Installing systemd-0:257~rc2- 100% |  59.7 MiB/s |  17.8 MiB |  00m00s
>>> Running post-install scriptlet: systemd-0:257~rc2-2.fc42.x86_64
>>> Finished post-install scriptlet: systemd-0:257~rc2-2.fc42.x86_64
>>> Scriptlet output:
>>> Creating group 'systemd-journal' with GID 190.
>>> Creating group 'systemd-oom' with GID 999.
>>> Creating user 'systemd-oom' (systemd Userspace OOM Killer) with UID 999 and 
>>> 
[ 73/216] Installing logrotate-0:3.22.0 100% |   3.2 MiB/s | 155.7 KiB |  00m00s
>>> Running post-install scriptlet: logrotate-0:3.22.0-2.fc41.x86_64
>>> Finished post-install scriptlet: logrotate-0:3.22.0-2.fc41.x86_64
>>> Scriptlet output:
>>> Created symlink '/etc/systemd/system/timers.target.wants/logrotate.timer' â†
>>> 
[ 74/216] Installing munge-0:0.5.16-3.f 100% |  13.8 MiB/s | 352.4 KiB |  00m00s
[ 75/216] Installing torque-libs-0:6.1. 100% | 112.1 MiB/s | 459.1 KiB |  00m00s
[ 76/216] Installing prrte-libs-0:3.0.6 100% | 153.4 MiB/s |   1.7 MiB |  00m00s
[ 77/216] Installing prrte-0:3.0.6-1.fc 100% |  58.2 MiB/s | 178.8 KiB |  00m00s
[ 78/216] Installing llvm18-static-0:18 100% | 379.0 MiB/s | 283.9 MiB |  00m01s
[ 79/216] Installing llvm18-googletest- 100% | 169.8 MiB/s |   2.2 MiB |  00m00s
[ 80/216] Installing tzdata-0:2024a-9.f 100% |  25.5 MiB/s |   1.9 MiB |  00m00s
[ 81/216] Installing python-pip-wheel-0 100% | 414.7 MiB/s |   1.2 MiB |  00m00s
[ 82/216] Installing mpdecimal-0:2.5.1- 100% | 100.6 MiB/s | 206.0 KiB |  00m00s
[ 83/216] Installing libb2-0:0.98.1-12. 100% |   7.0 MiB/s |  43.3 KiB |  00m00s
[ 84/216] Installing python3-libs-0:3.1 100% | 197.7 MiB/s |  40.7 MiB |  00m00s
[ 85/216] Installing python3-0:3.13.0-1 100% |  16.4 MiB/s |  33.5 KiB |  00m00s
[ 86/216] Installing llvm18-0:18.1.8-4. 100% | 307.5 MiB/s | 112.2 MiB |  00m00s
[ 87/216] Installing cmake-rpm-macros-0 100% |   7.9 MiB/s |   8.1 KiB |  00m00s
[ 88/216] Installing llvm18-test-0:18.1 100% | 149.7 MiB/s |   1.9 MiB |  00m00s
[ 89/216] Installing llvm18-devel-0:18. 100% | 148.1 MiB/s |  24.6 MiB |  00m00s
[ 90/216] Installing compiler-rt18-0:18 100% | 395.3 MiB/s |  28.1 MiB |  00m00s
[ 91/216] Installing clang18-tools-extr 100% | 342.7 MiB/s |  85.3 MiB |  00m00s
[ 92/216] Installing clang18-devel-0:18 100% | 253.5 MiB/s |  23.8 MiB |  00m00s
[ 93/216] Installing rocm-comgr-devel-0 100% | 102.0 MiB/s | 104.4 KiB |  00m00s
[ 94/216] Installing rocm-device-libs-0 100% | 266.9 MiB/s |   3.2 MiB |  00m00s
[ 95/216] Installing protobuf-c-0:1.5.0 100% |  54.2 MiB/s |  55.5 KiB |  00m00s
[ 96/216] Installing hiredis-0:1.2.0-4. 100% |   4.5 MiB/s | 115.7 KiB |  00m00s
[ 97/216] Installing unbound-libs-0:1.2 100% | 179.7 MiB/s |   1.4 MiB |  00m00s
[ 98/216] Installing gnutls-dane-0:3.8. 100% |  34.6 MiB/s |  70.8 KiB |  00m00s
[ 99/216] Installing ncurses-0:6.5-2.20 100% | 123.8 MiB/s | 633.9 KiB |  00m00s
[100/216] Installing perl-Digest-0:1.20 100% |  36.2 MiB/s |  37.1 KiB |  00m00s
[101/216] Installing perl-FileHandle-0: 100% |   9.5 MiB/s |   9.8 KiB |  00m00s
[102/216] Installing perl-B-0:1.89-512. 100% | 163.2 MiB/s | 501.3 KiB |  00m00s
[103/216] Installing perl-Digest-MD5-0: 100% |  60.2 MiB/s |  61.7 KiB |  00m00s
[104/216] Installing perl-MIME-Base32-0 100% |  31.4 MiB/s |  32.2 KiB |  00m00s
[105/216] Installing perl-Data-Dumper-0 100% |  55.5 MiB/s | 113.6 KiB |  00m00s
[106/216] Installing perl-libnet-0:3.15 100% |  95.9 MiB/s | 294.7 KiB |  00m00s
[107/216] Installing perl-IO-Socket-IP- 100% |  49.1 MiB/s | 100.5 KiB |  00m00s
[108/216] Installing perl-URI-0:5.31-1. 100% |  43.9 MiB/s | 269.6 KiB |  00m00s
[109/216] Installing perl-AutoLoader-0: 100% |   0.0   B/s |  20.9 KiB |  00m00s
[110/216] Installing perl-locale-0:1.12 100% |   0.0   B/s |   6.9 KiB |  00m00s
[111/216] Installing perl-Time-Local-2: 100% |  68.9 MiB/s |  70.6 KiB |  00m00s
[112/216] Installing perl-if-0:0.61.000 100% |   0.0   B/s |   6.2 KiB |  00m00s
[113/216] Installing perl-File-Path-0:2 100% |  63.0 MiB/s |  64.5 KiB |  00m00s
[114/216] Installing perl-Pod-Escapes-1 100% |  25.3 MiB/s |  25.9 KiB |  00m00s
[115/216] Installing perl-Text-Tabs+Wra 100% |  23.3 MiB/s |  23.9 KiB |  00m00s
[116/216] Installing perl-IO-Socket-SSL 100% | 172.7 MiB/s | 707.4 KiB |  00m00s
[117/216] Installing perl-Net-SSLeay-0: 100% | 136.3 MiB/s |   1.4 MiB |  00m00s
[118/216] Installing perl-POSIX-0:2.20- 100% | 115.4 MiB/s | 236.4 KiB |  00m00s
[119/216] Installing perl-Term-ANSIColo 100% |  96.9 MiB/s |  99.2 KiB |  00m00s
[120/216] Installing perl-Term-Cap-0:1. 100% |  29.9 MiB/s |  30.6 KiB |  00m00s
[121/216] Installing perl-IPC-Open3-0:1 100% |  22.7 MiB/s |  23.3 KiB |  00m00s
[122/216] Installing perl-Class-Struct- 100% |  25.3 MiB/s |  25.9 KiB |  00m00s
[123/216] Installing perl-File-Temp-1:0 100% | 160.2 MiB/s | 164.1 KiB |  00m00s
[124/216] Installing perl-HTTP-Tiny-0:0 100% |  76.4 MiB/s | 156.4 KiB |  00m00s
[125/216] Installing perl-Pod-Simple-1: 100% | 111.4 MiB/s | 570.5 KiB |  00m00s
[126/216] Installing perl-Symbol-0:1.09 100% |   0.0   B/s |   7.2 KiB |  00m00s
[127/216] Installing perl-SelectSaver-0 100% |   0.0   B/s |   2.6 KiB |  00m00s
[128/216] Installing perl-Socket-4:2.03 100% |  61.6 MiB/s | 126.1 KiB |  00m00s
[129/216] Installing perl-File-stat-0:1 100% |  12.7 MiB/s |  13.1 KiB |  00m00s
[130/216] Installing perl-Pod-Perldoc-0 100% |  82.6 MiB/s | 169.3 KiB |  00m00s
[131/216] Installing perl-podlators-1:6 100% | 104.6 MiB/s | 321.4 KiB |  00m00s
[132/216] Installing perl-Text-ParseWor 100% |  14.2 MiB/s |  14.6 KiB |  00m00s
[133/216] Installing perl-base-0:2.27-5 100% |   0.0   B/s |  12.9 KiB |  00m00s
[134/216] Installing perl-Fcntl-0:1.18- 100% |  48.9 MiB/s |  50.1 KiB |  00m00s
[135/216] Installing perl-mro-0:1.29-51 100% |   3.8 MiB/s |  46.7 KiB |  00m00s
[136/216] Installing perl-overloading-0 100% |   5.4 MiB/s |   5.5 KiB |  00m00s
[137/216] Installing perl-IO-0:1.55-512 100% |  75.8 MiB/s | 155.2 KiB |  00m00s
[138/216] Installing perl-Pod-Usage-4:2 100% |  84.3 MiB/s |  86.3 KiB |  00m00s
[139/216] Installing perl-Getopt-Std-0: 100% |   0.0   B/s |  11.7 KiB |  00m00s
[140/216] Installing perl-Scalar-List-U 100% |  49.7 MiB/s | 152.6 KiB |  00m00s
[141/216] Installing perl-constant-0:1. 100% |  26.7 MiB/s |  27.4 KiB |  00m00s
[142/216] Installing perl-Errno-0:1.38- 100% |   0.0   B/s |   8.8 KiB |  00m00s
[143/216] Installing perl-vars-0:1.05-5 100% |   0.0   B/s |   4.3 KiB |  00m00s
[144/216] Installing perl-MIME-Base64-0 100% |  23.6 MiB/s |  48.4 KiB |  00m00s
[145/216] Installing perl-parent-1:0.24 100% |  10.5 MiB/s |  10.7 KiB |  00m00s
[146/216] Installing perl-overload-0:1. 100% |  70.3 MiB/s |  71.9 KiB |  00m00s
[147/216] Installing perl-Storable-1:3. 100% | 114.3 MiB/s | 234.0 KiB |  00m00s
[148/216] Installing perl-Getopt-Long-1 100% |  71.9 MiB/s | 147.2 KiB |  00m00s
[149/216] Installing perl-File-Basename 100% |   0.0   B/s |  14.6 KiB |  00m00s
[150/216] Installing perl-Carp-0:1.54-5 100% |  46.6 MiB/s |  47.7 KiB |  00m00s
[151/216] Installing perl-Exporter-0:5. 100% |  54.3 MiB/s |  55.6 KiB |  00m00s
[152/216] Installing perl-PathTools-0:3 100% |  60.1 MiB/s | 184.6 KiB |  00m00s
[153/216] Installing perl-DynaLoader-0: 100% |  31.7 MiB/s |  32.5 KiB |  00m00s
[154/216] Installing perl-Encode-4:3.21 100% | 224.7 MiB/s |   4.7 MiB |  00m00s
[155/216] Installing perl-libs-4:5.40.0 100% | 140.5 MiB/s |  10.0 MiB |  00m00s
[156/216] Installing perl-interpreter-4 100% | 121.1 MiB/s | 124.0 KiB |  00m00s
[157/216] Installing hipcc-0:18-25.rocm 100% | 215.4 MiB/s | 661.8 KiB |  00m00s
[158/216] Installing perl-File-Find-0:1 100% |  41.5 MiB/s |  42.5 KiB |  00m00s
[159/216] Installing perl-TermReadKey-0 100% |  32.4 MiB/s |  66.3 KiB |  00m00s
[160/216] Installing perl-lib-0:0.65-51 100% |   0.0   B/s |   8.9 KiB |  00m00s
[161/216] Installing perl-File-Copy-0:2 100% |   0.0   B/s |  20.2 KiB |  00m00s
[162/216] Installing perl-File-Which-0: 100% |  30.7 MiB/s |  31.4 KiB |  00m00s
[163/216] Installing perl-Error-1:0.170 100% |  39.3 MiB/s |  80.5 KiB |  00m00s
[164/216] Installing hwdata-0:0.389-1.f 100% | 387.5 MiB/s |   9.3 MiB |  00m00s
[165/216] Installing libpciaccess-0:0.1 100% |  44.9 MiB/s |  46.0 KiB |  00m00s
[166/216] Installing libdrm-0:2.4.123-1 100% | 134.1 MiB/s | 411.9 KiB |  00m00s
[167/216] Installing hsakmt-0:1.0.6-49. 100% | 180.5 MiB/s | 184.8 KiB |  00m00s
[168/216] Installing rocm-runtime-0:6.2 100% | 337.1 MiB/s |   2.7 MiB |  00m00s
[169/216] Installing rocm-hip-0:6.2.1-1 100% | 291.7 MiB/s |  23.0 MiB |  00m00s
[170/216] Installing hsakmt-devel-0:1.0 100% | 109.9 MiB/s | 112.6 KiB |  00m00s
[171/216] Installing rocm-runtime-devel 100% |  19.5 MiB/s | 559.9 KiB |  00m00s
[172/216] Installing tpm2-tss-0:4.1.3-3 100% | 158.1 MiB/s |   1.6 MiB |  00m00s
[173/216] Installing npth-0:1.8-1.fc42. 100% |  53.4 MiB/s |  54.7 KiB |  00m00s
[174/216] Installing gnupg2-0:2.4.5-4.f 100% | 223.0 MiB/s |   9.6 MiB |  00m00s
[175/216] Installing gpgme-0:1.24.0-1.f 100% | 143.9 MiB/s | 589.3 KiB |  00m00s
[176/216] Installing wget2-libs-0:2.1.0 100% | 179.2 MiB/s | 366.9 KiB |  00m00s
[177/216] Installing wget2-0:2.1.0-12.f 100% | 144.8 MiB/s | 889.7 KiB |  00m00s
[178/216] Installing libpipeline-0:1.5. 100% |   7.4 MiB/s | 150.6 KiB |  00m00s
[179/216] Installing man-db-0:2.13.0-1. 100% |  67.3 MiB/s |   2.9 MiB |  00m00s
[180/216] Installing environment-module 100% |  62.2 MiB/s |   1.8 MiB |  00m00s
[181/216] Installing rocm-rpm-macros-mo 100% |   3.8 MiB/s |  31.5 KiB |  00m00s
[182/216] Installing rocblas-0:6.2.4-4. 100% | 529.4 MiB/s |   3.7 GiB |  00m07s
[183/216] Installing rocsparse-0:6.2.1- 100% | 432.6 MiB/s | 860.0 MiB |  00m02s
[184/216] Installing rocsolver-0:6.2.4- 100% | 441.2 MiB/s | 445.6 MiB |  00m01s
[185/216] Installing hipblas-0:6.2.0-8. 100% | 325.4 MiB/s |   1.0 MiB |  00m00s
[186/216] Installing libcbor-0:0.11.0-2 100% |  73.5 MiB/s |  75.3 KiB |  00m00s
[187/216] Installing libfido2-0:1.15.0- 100% | 117.1 MiB/s | 239.7 KiB |  00m00s
[188/216] Installing tcsh-0:6.24.13-1.f 100% |  51.9 MiB/s |   1.2 MiB |  00m00s
[189/216] Installing orangefs-0:2.9.8-1 100% | 163.7 MiB/s |   3.1 MiB |  00m00s
[190/216] Installing openssh-0:9.9p1-5. 100% | 229.9 MiB/s |   1.4 MiB |  00m00s
[191/216] Installing openssh-clients-0: 100% |  70.0 MiB/s |   2.7 MiB |  00m00s
[192/216] Installing git-core-0:2.47.1- 100% | 279.7 MiB/s |  22.7 MiB |  00m00s
[193/216] Installing git-core-doc-0:2.4 100% | 187.4 MiB/s |  17.4 MiB |  00m00s
[194/216] Installing perl-Git-0:2.47.1- 100% |  63.5 MiB/s |  65.0 KiB |  00m00s
[195/216] Installing git-0:2.47.1-1.fc4 100% |  85.4 MiB/s |  87.4 KiB |  00m00s
[196/216] Installing pthreadpool-0:0.0^ 100% | 111.6 MiB/s | 114.3 KiB |  00m00s
[197/216] Installing rhash-0:1.4.5-1.fc 100% | 118.7 MiB/s | 364.6 KiB |  00m00s
[198/216] Installing libuv-1:1.49.2-1.f 100% |  46.6 MiB/s | 572.2 KiB |  00m00s
[199/216] Installing cmake-data-0:3.30. 100% |  56.0 MiB/s |   8.8 MiB |  00m00s
[200/216] Installing cmake-0:3.30.5-1.f 100% | 279.4 MiB/s |  32.7 MiB |  00m00s
[201/216] Installing ucx-0:1.17.0-3.fc4 100% | 215.1 MiB/s |   2.4 MiB |  00m00s
[202/216] Installing libquadmath-0:14.2 100% | 159.8 MiB/s | 327.2 KiB |  00m00s
[203/216] Installing libgfortran-0:14.2 100% | 304.5 MiB/s |   3.0 MiB |  00m00s
[204/216] Installing openmpi-0:5.0.6-1. 100% | 250.8 MiB/s |   7.0 MiB |  00m00s
[205/216] Installing pthreadpool-devel- 100% |  97.5 MiB/s |  99.8 KiB |  00m00s
[206/216] Installing hipblas-devel-0:6. 100% | 449.3 MiB/s |   2.7 MiB |  00m00s
[207/216] Installing rocblas-devel-0:6. 100% | 407.3 MiB/s |   2.4 MiB |  00m00s
[208/216] Installing rocm-rpm-macros-0: 100% |  19.2 MiB/s |  19.6 KiB |  00m00s
[209/216] Installing wget2-wget-0:2.1.0 100% | 433.6 KiB/s | 444.0   B |  00m00s
[210/216] Installing rocm-hip-devel-0:6 100% | 215.5 MiB/s |   2.6 MiB |  00m00s
[211/216] Installing libcurl-devel-0:8. 100% |  46.8 MiB/s |   1.4 MiB |  00m00s
[212/216] Installing annobin-plugin-gcc 100% |  37.3 MiB/s | 992.9 KiB |  00m00s
[213/216] Installing gcc-plugin-annobin 100% |   2.2 MiB/s |  59.2 KiB |  00m00s
[214/216] Installing langpacks-en-0:4.2 100% | 683.6 KiB/s | 700.0   B |  00m00s
[215/216] Installing systemd-rpm-macros 100% |  11.0 MiB/s |  11.2 KiB |  00m00s
[216/216] Installing xxd-2:9.1.866-1.fc 100% | 119.6 KiB/s |  47.5 KiB |  00m00s
Warning: skipped PGP checks for 16 packages from repository: copr_base
Complete!
Finish: build setup for llama-cpp-b4094-6.fc42.src.rpm
Start: rpmbuild llama-cpp-b4094-6.fc42.src.rpm
Building target platforms: x86_64
Building for target x86_64
setting SOURCE_DATE_EPOCH=1732665600
Executing(%mkbuilddir): /bin/sh -e /var/tmp/rpm-tmp.6v1sj4
+ umask 022
+ cd /builddir/build/BUILD/llama-cpp-b4094-build
+ test -d /builddir/build/BUILD/llama-cpp-b4094-build
+ /usr/bin/chmod -Rf a+rX,u+w,g-w,o-w /builddir/build/BUILD/llama-cpp-b4094-build
+ /usr/bin/rm -rf /builddir/build/BUILD/llama-cpp-b4094-build
+ /usr/bin/mkdir -p /builddir/build/BUILD/llama-cpp-b4094-build
+ /usr/bin/mkdir -p /builddir/build/BUILD/llama-cpp-b4094-build/SPECPARTS
+ RPM_EC=0
++ jobs -p
+ exit 0
Executing(%prep): /bin/sh -e /var/tmp/rpm-tmp.sxjo0h
+ umask 022
+ cd /builddir/build/BUILD/llama-cpp-b4094-build
+ cd /builddir/build/BUILD/llama-cpp-b4094-build
+ rm -rf llama.cpp-b4094
+ /usr/lib/rpm/rpmuncompress -x /builddir/build/SOURCES/llama.cpp-b4094.tar.gz
+ STATUS=0
+ '[' 0 -ne 0 ']'
+ cd llama.cpp-b4094
+ /usr/bin/chmod -Rf a+rX,u+w,g-w,o-w .
+ sed -i -e 's/POSITION_INDEPENDENT_CODE ON/POSITION_INDEPENDENT_CODE ON SOVERSION b4094/' src/CMakeLists.txt
+ sed -i -e 's/POSITION_INDEPENDENT_CODE ON/POSITION_INDEPENDENT_CODE ON SOVERSION b4094/' ggml/src/CMakeLists.txt
+ rm -rf exmples/llma.android
+ find . -name .gitignore -exec rm -rf '{}' ';'
+ RPM_EC=0
++ jobs -p
+ exit 0
Executing(%build): /bin/sh -e /var/tmp/rpm-tmp.hsh3UO
+ umask 022
+ cd /builddir/build/BUILD/llama-cpp-b4094-build
+ CFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer '
+ export CFLAGS
+ CXXFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer'
+ export CXXFLAGS
+ FFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -I/usr/lib64/gfortran/modules '
+ export FFLAGS
+ FCFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -I/usr/lib64/gfortran/modules '
+ export FCFLAGS
+ VALAFLAGS=-g
+ export VALAFLAGS
+ RUSTFLAGS='-Copt-level=3 -Cdebuginfo=2 -Ccodegen-units=1 -Cstrip=none -Cforce-frame-pointers=yes -Clink-arg=-specs=/usr/lib/rpm/redhat/redhat-package-notes --cap-lints=warn'
+ export RUSTFLAGS
+ LDFLAGS='-Wl,-z,relro -Wl,--as-needed  -Wl,-z,pack-relative-relocs -Wl,-z,now -Wl,-z,now   -Wl,--build-id=sha1 -specs=/usr/lib/rpm/redhat/redhat-package-notes '
+ export LDFLAGS
+ LT_SYS_LIBRARY_PATH=/usr/lib64:
+ export LT_SYS_LIBRARY_PATH
+ CC=hipcc
+ export CC
+ CXX=hipcc
+ export CXX
+ cd llama.cpp-b4094
+ module load rocm/default
+ local _mlredir=0
+ '[' -n '' ']'
+ case " $@ " in
+ '[' 0 -eq 0 ']'
+ _module_raw load rocm/default
++ /usr/bin/tclsh /usr/share/Modules/libexec/modulecmd.tcl bash load rocm/default
+ eval '__MODULES_LMCONFLICT=rocm/default\&rocm; export __MODULES_LMCONFLICT;
ROCM_BIN=/usr/bin; export ROCM_BIN;
_LMFILES_=/usr/share/modulefiles/rocm/default; export _LMFILES_;
LOADEDMODULES=rocm/default; export LOADEDMODULES;
ROCM_GPUS=gfx900\;gfx906:xnack-\;gfx908:xnack-\;gfx90a:xnack+\;gfx90a:xnack-\;gfx942\;gfx1010\;gfx1012\;gfx1030\;gfx1031\;gfx1035\;gfx1100\;gfx1101\;gfx1102\;gfx1103\;gfx1151; export ROCM_GPUS;
ROCM_LIB=/usr/lib64; export ROCM_LIB;
test 0;'
++ __MODULES_LMCONFLICT='rocm/default&rocm'
++ export __MODULES_LMCONFLICT
++ ROCM_BIN=/usr/bin
++ export ROCM_BIN
++ _LMFILES_=/usr/share/modulefiles/rocm/default
++ export _LMFILES_
++ LOADEDMODULES=rocm/default
++ export LOADEDMODULES
++ ROCM_GPUS='gfx900;gfx906:xnack-;gfx908:xnack-;gfx90a:xnack+;gfx90a:xnack-;gfx942;gfx1010;gfx1012;gfx1030;gfx1031;gfx1035;gfx1100;gfx1101;gfx1102;gfx1103;gfx1151'
++ export ROCM_GPUS
++ ROCM_LIB=/usr/lib64
++ export ROCM_LIB
++ test 0
+ _mlstatus=0
+ return 0
+ CFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer '
+ export CFLAGS
+ CXXFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer'
+ export CXXFLAGS
+ FFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -I/usr/lib64/gfortran/modules '
+ export FFLAGS
+ FCFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -I/usr/lib64/gfortran/modules '
+ export FCFLAGS
+ VALAFLAGS=-g
+ export VALAFLAGS
+ RUSTFLAGS='-Copt-level=3 -Cdebuginfo=2 -Ccodegen-units=1 -Cstrip=none -Cforce-frame-pointers=yes -Clink-arg=-specs=/usr/lib/rpm/redhat/redhat-package-notes --cap-lints=warn'
+ export RUSTFLAGS
+ LDFLAGS='-Wl,-z,relro -Wl,--as-needed  -Wl,-z,pack-relative-relocs -Wl,-z,now -Wl,-z,now   -Wl,--build-id=sha1 -specs=/usr/lib/rpm/redhat/redhat-package-notes '
+ export LDFLAGS
+ LT_SYS_LIBRARY_PATH=/usr/lib64:
+ export LT_SYS_LIBRARY_PATH
+ CC=hipcc
+ export CC
+ CXX=hipcc
+ export CXX
+ /usr/bin/cmake -S . -B redhat-linux-build -DCMAKE_C_FLAGS_RELEASE:STRING=-DNDEBUG -DCMAKE_CXX_FLAGS_RELEASE:STRING=-DNDEBUG -DCMAKE_Fortran_FLAGS_RELEASE:STRING=-DNDEBUG -DCMAKE_VERBOSE_MAKEFILE:BOOL=ON -DCMAKE_INSTALL_DO_STRIP:BOOL=OFF -DCMAKE_INSTALL_PREFIX:PATH=/usr -DINCLUDE_INSTALL_DIR:PATH=/usr/include -DLIB_INSTALL_DIR:PATH=/usr/lib64 -DSYSCONF_INSTALL_DIR:PATH=/etc -DSHARE_INSTALL_PREFIX:PATH=/usr/share -DLIB_SUFFIX=64 -DBUILD_SHARED_LIBS:BOOL=ON -DCMAKE_INSTALL_LIBDIR=lib64 -DCMAKE_SKIP_RPATH=ON -DGGML_AVX=OFF -DGGML_AVX2=OFF -DGGML_AVX512=OFF -DGGML_AVX512_VBMI=OFF -DGGML_AVX512_VNNI=OFF -DGGML_FMA=OFF -DGGML_F16C=OFF -DGGML_HIP=ON '-DAMDGPU_TARGETS=gfx900;gfx906:xnack-;gfx908:xnack-;gfx90a:xnack+;gfx90a:xnack-;gfx942;gfx1010;gfx1012;gfx1030;gfx1031;gfx1035;gfx1100;gfx1101;gfx1102;gfx1103;gfx1151' -DLLAMA_BUILD_EXAMPLES=OFF -DLLAMA_BUILD_TESTS=OFF
-- The C compiler identification is Clang 18.1.8
-- The CXX compiler identification is Clang 18.1.8
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: /usr/bin/hipcc - skipped
-- Detecting C compile features
-- Detecting C compile features - done
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Check for working CXX compiler: /usr/bin/hipcc - skipped
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Found Git: /usr/bin/git (found version "2.47.1")
fatal: not a git repository (or any of the parent directories): .git
fatal: not a git repository (or any of the parent directories): .git
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD - Success
-- Found Threads: TRUE
-- Warning: ccache not found - consider installing it for faster compilation or disable this warning with GGML_CCACHE=OFF
-- CMAKE_SYSTEM_PROCESSOR: x86_64
-- Could NOT find OpenMP_C (missing: OpenMP_C_FLAGS OpenMP_C_LIB_NAMES) 
-- Could NOT find OpenMP_CXX (missing: OpenMP_CXX_FLAGS OpenMP_CXX_LIB_NAMES) 
-- Could NOT find OpenMP (missing: OpenMP_C_FOUND OpenMP_CXX_FOUND) 
-- Using llamafile
-- x86 detected
-- Using runtime weight conversion of Q4_0 to Q4_0_x_x to enable optimized GEMM/GEMV kernels
CMake Warning at ggml/src/ggml-cpu/CMakeLists.txt:43 (message):
  OpenMP not found


-- Including CPU backend
CMake Warning at ggml/src/ggml-amx/CMakeLists.txt:106 (message):
  AMX requires x86 and gcc version > 11.0.  Turning off GGML_AMX.


CMake Warning at ggml/src/ggml-hip/CMakeLists.txt:27 (message):
  Setting hipcc as the C++ compiler is legacy behavior.  Prefer setting the
  HIP compiler directly.  See README for details.


-- Performing Test HIP_CLANG_SUPPORTS_PARALLEL_JOBS
-- Performing Test HIP_CLANG_SUPPORTS_PARALLEL_JOBS - Failed
-- HIP and hipBLAS found
-- Including HIP backend
CMake Warning at common/CMakeLists.txt:30 (message):
  Git repository not found; to enable automatic generation of build info,
  make sure Git is installed and the project is a Git repository.


-- Configuring done (8.0s)
-- Generating done (0.0s)
CMake Warning:
  Manually-specified variables were not used by the project:

    CMAKE_Fortran_FLAGS_RELEASE
    CMAKE_INSTALL_D-- Build files have been written to: /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build
O_STRIP
    INCLUDE_INSTALL_DIR
    LIB_INSTALL_DIR
    LIB_SUFFIX
    SHARE_INSTALL_PREFIX
    SYSCONF_INSTALL_DIR


+ /usr/bin/cmake --build redhat-linux-build -j2 --verbose
Change Dir: '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'

Run Build Command(s): /usr/bin/cmake -E env VERBOSE=1 /usr/bin/gmake -f Makefile -j2
/usr/bin/cmake -S/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094 -B/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build --check-build-system CMakeFiles/Makefile.cmake 0
/usr/bin/cmake -E cmake_progress_start /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/CMakeFiles /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build//CMakeFiles/progress.marks
/usr/bin/gmake  -f CMakeFiles/Makefile2 all
gmake[1]: Entering directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
/usr/bin/gmake  -f ggml/src/CMakeFiles/ggml-base.dir/build.make ggml/src/CMakeFiles/ggml-base.dir/depend
/usr/bin/gmake  -f common/CMakeFiles/build_info.dir/build.make common/CMakeFiles/build_info.dir/depend
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build && /usr/bin/cmake -E cmake_depends "Unix Makefiles" /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094 /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/CMakeFiles/ggml-base.dir/DependInfo.cmake "--color="
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
/usr/bin/gmake  -f ggml/src/CMakeFiles/ggml-base.dir/build.make ggml/src/CMakeFiles/ggml-base.dir/build
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
[  0%] Generating build details from Git
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094 && /usr/bin/cmake -DMSVC= -DCMAKE_C_COMPILER_VERSION=18.1.8 -DCMAKE_C_COMPILER_ID=Clang -DCMAKE_VS_PLATFORM_NAME= -DCMAKE_C_COMPILER=/usr/bin/hipcc -P /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/cmake/build-info-gen-cpp.cmake
[  1%] Building C object ggml/src/CMakeFiles/ggml-base.dir/ggml.c.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu11 -fPIC -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes -Werror=implicit-int -Werror=implicit-function-declaration -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wdouble-promotion -MD -MT ggml/src/CMakeFiles/ggml-base.dir/ggml.c.o -MF CMakeFiles/ggml-base.dir/ggml.c.o.d -o CMakeFiles/ggml-base.dir/ggml.c.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml.c
-- Found Git: /usr/bin/git (found version "2.47.1")
fatal: not a git repository (or any of the parent directories): .git
fatal: not a git repository (or any of the parent directories): .git
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build && /usr/bin/cmake -E cmake_depends "Unix Makefiles" /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094 /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/common /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/common/CMakeFiles/build_info.dir/DependInfo.cmake "--color="
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
/usr/bin/gmake  -f common/CMakeFiles/build_info.dir/build.make common/CMakeFiles/build_info.dir/build
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
[  2%] Building CXX object common/CMakeFiles/build_info.dir/build-info.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/common && /usr/bin/hipcc   -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -MD -MT common/CMakeFiles/build_info.dir/build-info.cpp.o -MF CMakeFiles/build_info.dir/build-info.cpp.o.d -o CMakeFiles/build_info.dir/build-info.cpp.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/build-info.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[  3%] Building C object ggml/src/CMakeFiles/ggml-base.dir/ggml-alloc.c.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu11 -fPIC -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes -Werror=implicit-int -Werror=implicit-function-declaration -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wdouble-promotion -MD -MT ggml/src/CMakeFiles/ggml-base.dir/ggml-alloc.c.o -MF CMakeFiles/ggml-base.dir/ggml-alloc.c.o.d -o CMakeFiles/ggml-base.dir/ggml-alloc.c.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-alloc.c
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[  4%] Building CXX object ggml/src/CMakeFiles/ggml-base.dir/ggml-backend.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT ggml/src/CMakeFiles/ggml-base.dir/ggml-backend.cpp.o -MF CMakeFiles/ggml-base.dir/ggml-backend.cpp.o.d -o CMakeFiles/ggml-base.dir/ggml-backend.cpp.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-backend.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
[  4%] Built target build_info
[  5%] Building CXX object ggml/src/CMakeFiles/ggml-base.dir/ggml-threading.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT ggml/src/CMakeFiles/ggml-base.dir/ggml-threading.cpp.o -MF CMakeFiles/ggml-base.dir/ggml-threading.cpp.o.d -o CMakeFiles/ggml-base.dir/ggml-threading.cpp.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-threading.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[  6%] Building C object ggml/src/CMakeFiles/ggml-base.dir/ggml-quants.c.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu11 -fPIC -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes -Werror=implicit-int -Werror=implicit-function-declaration -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wdouble-promotion -MD -MT ggml/src/CMakeFiles/ggml-base.dir/ggml-quants.c.o -MF CMakeFiles/ggml-base.dir/ggml-quants.c.o.d -o CMakeFiles/ggml-base.dir/ggml-quants.c.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-quants.c
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[  7%] Building C object ggml/src/CMakeFiles/ggml-base.dir/ggml-aarch64.c.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu11 -fPIC -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes -Werror=implicit-int -Werror=implicit-function-declaration -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wdouble-promotion -MD -MT ggml/src/CMakeFiles/ggml-base.dir/ggml-aarch64.c.o -MF CMakeFiles/ggml-base.dir/ggml-aarch64.c.o.d -o CMakeFiles/ggml-base.dir/ggml-aarch64.c.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-aarch64.c
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[  8%] Linking CXX shared library libggml-base.so
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src && /usr/bin/cmake -E cmake_link_script CMakeFiles/ggml-base.dir/link.txt --verbose=1
/usr/bin/hipcc -fPIC -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -Wl,-z,relro -Wl,--as-needed  -Wl,-z,pack-relative-relocs -Wl,-z,now -Wl,-z,now   -Wl,--build-id=sha1 -specs=/usr/lib/rpm/redhat/redhat-package-notes -shared -Wl,-soname,libggml-base.so.b4094 -o libggml-base.so.b4094 "CMakeFiles/ggml-base.dir/ggml.c.o" "CMakeFiles/ggml-base.dir/ggml-alloc.c.o" "CMakeFiles/ggml-base.dir/ggml-backend.cpp.o" "CMakeFiles/ggml-base.dir/ggml-threading.cpp.o" "CMakeFiles/ggml-base.dir/ggml-quants.c.o" "CMakeFiles/ggml-base.dir/ggml-aarch64.c.o"  -lm
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
clang++: warning: argument unused during compilation: '-Xarch_host -fstack-protector-strong' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-Xarch_host -fcf-protection' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src && /usr/bin/cmake -E cmake_symlink_library libggml-base.so.b4094 libggml-base.so.b4094 libggml-base.so
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
[  8%] Built target ggml-base
/usr/bin/gmake  -f ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/build.make ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/depend
/usr/bin/gmake  -f ggml/src/ggml-cpu/CMakeFiles/ggml-cpu.dir/build.make ggml/src/ggml-cpu/CMakeFiles/ggml-cpu.dir/depend
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build && /usr/bin/cmake -E cmake_depends "Unix Makefiles" /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094 /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cpu /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-cpu /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-cpu/CMakeFiles/ggml-cpu.dir/DependInfo.cmake "--color="
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build && /usr/bin/cmake -E cmake_depends "Unix Makefiles" /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094 /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/DependInfo.cmake "--color="
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
/usr/bin/gmake  -f ggml/src/ggml-cpu/CMakeFiles/ggml-cpu.dir/build.make ggml/src/ggml-cpu/CMakeFiles/ggml-cpu.dir/build
/usr/bin/gmake  -f ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/build.make ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/build
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
[  9%] Building C object ggml/src/ggml-cpu/CMakeFiles/ggml-cpu.dir/ggml-cpu.c.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-cpu && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_AARCH64 -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cpu/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cpu/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu11 -fPIC -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes -Werror=implicit-int -Werror=implicit-function-declaration -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wdouble-promotion -march=native -MD -MT ggml/src/ggml-cpu/CMakeFiles/ggml-cpu.dir/ggml-cpu.c.o -MF CMakeFiles/ggml-cpu.dir/ggml-cpu.c.o.d -o CMakeFiles/ggml-cpu.dir/ggml-cpu.c.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cpu/ggml-cpu.c
[ 10%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/acc.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/acc.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/acc.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/acc.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cu
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
[ 11%] Building CXX object ggml/src/ggml-cpu/CMakeFiles/ggml-cpu.dir/ggml-cpu.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-cpu && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_AARCH64 -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cpu/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cpu/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -march=native -MD -MT ggml/src/ggml-cpu/CMakeFiles/ggml-cpu.dir/ggml-cpu.cpp.o -MF CMakeFiles/ggml-cpu.dir/ggml-cpu.cpp.o.d -o CMakeFiles/ggml-cpu.dir/ggml-cpu.cpp.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cpu/ggml-cpu.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
[ 12%] Building C object ggml/src/ggml-cpu/CMakeFiles/ggml-cpu.dir/ggml-cpu-aarch64.c.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-cpu && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_AARCH64 -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cpu/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cpu/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu11 -fPIC -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes -Werror=implicit-int -Werror=implicit-function-declaration -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wdouble-promotion -march=native -MD -MT ggml/src/ggml-cpu/CMakeFiles/ggml-cpu.dir/ggml-cpu-aarch64.c.o -MF CMakeFiles/ggml-cpu.dir/ggml-cpu-aarch64.c.o.d -o CMakeFiles/ggml-cpu.dir/ggml-cpu-aarch64.c.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cpu/ggml-cpu-aarch64.c
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
[ 13%] Building C object ggml/src/ggml-cpu/CMakeFiles/ggml-cpu.dir/ggml-cpu-quants.c.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-cpu && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_AARCH64 -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cpu/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cpu/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu11 -fPIC -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes -Werror=implicit-int -Werror=implicit-function-declaration -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wdouble-promotion -march=native -MD -MT ggml/src/ggml-cpu/CMakeFiles/ggml-cpu.dir/ggml-cpu-quants.c.o -MF CMakeFiles/ggml-cpu.dir/ggml-cpu-quants.c.o.d -o CMakeFiles/ggml-cpu.dir/ggml-cpu-quants.c.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cpu/ggml-cpu-quants.c
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
6 warnings generated when compiling for gfx1035.
[ 14%] Building CXX object ggml/src/ggml-cpu/CMakeFiles/ggml-cpu.dir/llamafile/sgemm.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-cpu && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_AARCH64 -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cpu/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cpu/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -march=native -MD -MT ggml/src/ggml-cpu/CMakeFiles/ggml-cpu.dir/llamafile/sgemm.cpp.o -MF CMakeFiles/ggml-cpu.dir/llamafile/sgemm.cpp.o.d -o CMakeFiles/ggml-cpu.dir/llamafile/sgemm.cpp.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cpu/llamafile/sgemm.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
[ 15%] Linking CXX shared library libggml-cpu.so
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-cpu && /usr/bin/cmake -E cmake_link_script CMakeFiles/ggml-cpu.dir/link.txt --verbose=1
/usr/bin/hipcc -fPIC -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -Wl,-z,relro -Wl,--as-needed  -Wl,-z,pack-relative-relocs -Wl,-z,now -Wl,-z,now   -Wl,--build-id=sha1 -specs=/usr/lib/rpm/redhat/redhat-package-notes -shared -Wl,-soname,libggml-cpu.so -o libggml-cpu.so "CMakeFiles/ggml-cpu.dir/ggml-cpu.c.o" "CMakeFiles/ggml-cpu.dir/ggml-cpu.cpp.o" "CMakeFiles/ggml-cpu.dir/ggml-cpu-aarch64.c.o" "CMakeFiles/ggml-cpu.dir/ggml-cpu-quants.c.o" "CMakeFiles/ggml-cpu.dir/llamafile/sgemm.cpp.o"  ../libggml-base.so.b4094
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
clang++: warning: argument unused during compilation: '-Xarch_host -fstack-protector-strong' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-Xarch_host -fcf-protection' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
[ 15%] Built target ggml-cpu
[ 16%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/arange.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/arange.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/arange.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/arange.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 17%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/argmax.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/argmax.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/argmax.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/argmax.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argmax.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 18%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/argsort.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/argsort.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/argsort.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/argsort.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cu
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 19%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/binbcast.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/binbcast.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/binbcast.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/binbcast.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:353:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  353 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
7 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:353:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  353 |         } break;
      |           ^~~~~
7 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:353:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  353 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
7 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:353:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  353 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
7 warnings generated when compiling for gfx1031.
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:353:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  353 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
7 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 20%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/clamp.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/clamp.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/clamp.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/clamp.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cu
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:353:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  353 |         } break;
      |           ^~~~~
7 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:353:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  353 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
7 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:353:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  353 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
7 warnings generated when compiling for gfx1102.
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:353:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  353 |         } break;
      |           ^~~~~
6 warnings generated when compiling for gfx1035.
7 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:353:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  353 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
7 warnings generated when compiling for gfx1151.
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:353:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  353 |         } break;
      |           ^~~~~
6 warnings generated when compiling for gfx1102.
7 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:353:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  353 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
7 warnings generated when compiling for gfx906.
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:353:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  353 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
7 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:353:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  353 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
7 warnings generated when compiling for gfx90a.
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:353:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  353 |         } break;
      |           ^~~~~
6 warnings generated when compiling for gfx90a.
7 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:353:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  353 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
7 warnings generated when compiling for gfx942.
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/binbcast.cu:353:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  353 |         } break;
      |           ^~~~~
6 warnings generated when compiling for host.
[ 21%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/concat.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/concat.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/concat.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/concat.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu
7 warnings generated when compiling for host.
[ 22%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv-transpose-1d.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv-transpose-1d.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv-transpose-1d.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv-transpose-1d.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
8 warnings generated when compiling for gfx1010.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
8 warnings generated when compiling for gfx1012.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
8 warnings generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
8 warnings generated when compiling for gfx1031.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu::41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
 warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
8 warnings generated when compiling for gfx1035.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
8 warnings generated when compiling for gfx1101.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
8 warnings generated when compiling for gfx1102.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1103.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1151.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx900.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx906.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx908.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx942.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for host.
[ 23%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/convert.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/convert.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/convert.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/convert.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for host.
[ 23%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/count-equal.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/count-equal.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/count-equal.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/count-equal.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1010.
12 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
12 warnings generated when compiling for gfx1012.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1031.
12 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
12 warnings generated when compiling for gfx1031.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1101.
12 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx1100.
7 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
7 warnings generated when compiling for gfx1151.
12 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx906.
12 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx908.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
12 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx90a.
12 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx942.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
12 warnings generated when compiling for gfx900.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for host.
[ 24%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cpy.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cpy.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/cpy.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/cpy.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
6 warnings generated when compiling for gfx1010.
12 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
6 warnings generated when compiling for gfx1012.
12 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
6 warnings generated when compiling for gfx1030.
12 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
6 warnings generated when compiling for gfx1031.
12 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
6 warnings generated when compiling for gfx1035.
12 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
6 warnings generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
12 warnings generated when compiling for host.
[ 25%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cross-entropy-loss.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cross-entropy-loss.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/cross-entropy-loss.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/cross-entropy-loss.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cross-entropy-loss.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 26%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/diagmask.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/diagmask.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/diagmask.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/diagmask.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cu
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
6 warnings generated when compiling for host.
[ 27%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/dmmv.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/dmmv.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/dmmv.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/dmmv.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:420:37: warning: cast from 'const __half *' to '__half2 *' drops const qualifier [-Wcast-qual]
  420 |     const half2 x_reg = *((half2 *) &(x[ib + iqs]));
      |                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:682:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  682 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:524:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_0>' requested here
  524 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:533:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_1>' requested here
  533 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:542:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_0>' requested here
  542 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:551:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_1>' requested here
  551 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:560:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q8_0>' requested here
  560 |     dequantize_mul_mat_vec<GGML_TYPE_Q8_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:611:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_F16>' requested here
  611 |     dequantize_mul_mat_vec<GGML_TYPE_F16>
      |     ^
15 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:420:37: warning: cast from 'const __half *' to '__half2 *' drops const qualifier [-Wcast-qual]
  420 |     const half2 x_reg = *((half2 *) &(x[ib + iqs]));
      |                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:682:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  682 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:524:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_0>' requested here
  524 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:533:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_1>' requested here
  533 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:542:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_0>' requested here
  542 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:551:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_1>' requested here
  551 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:560:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q8_0>' requested here
  560 |     dequantize_mul_mat_vec<GGML_TYPE_Q8_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:611:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_F16>' requested here
  611 |     dequantize_mul_mat_vec<GGML_TYPE_F16>
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
15 warnings generated when compiling for gfx1012.
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:420:37: warning: cast from 'const __half *' to '__half2 *' drops const qualifier [-Wcast-qual]
  420 |     const half2 x_reg = *((half2 *) &(x[ib + iqs]));
      |                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:682:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  682 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:524:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_0>' requested here
  524 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:533:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_1>' requested here
  533 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:542:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_0>' requested here
  542 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:551:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_1>' requested here
  551 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:560:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q8_0>' requested here
  560 |     dequantize_mul_mat_vec<GGML_TYPE_Q8_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:611:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_F16>' requested here
  611 |     dequantize_mul_mat_vec<GGML_TYPE_F16>
      |     ^
6 warnings generated when compiling for gfx1100.
15 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:420:37: warning: cast from 'const __half *' to '__half2 *' drops const qualifier [-Wcast-qual]
  420 |     const half2 x_reg = *((half2 *) &(x[ib + iqs]));
      |                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:682:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  682 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:524:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_0>' requested here
  524 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:533:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_1>' requested here
  533 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:542:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_0>' requested here
  542 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:551:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_1>' requested here
  551 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:560:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q8_0>' requested here
  560 |     dequantize_mul_mat_vec<GGML_TYPE_Q8_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:611:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_F16>' requested here
  611 |     dequantize_mul_mat_vec<GGML_TYPE_F16>
      |     ^
15 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:420:37: warning: cast from 'const __half *' to '__half2 *' drops const qualifier [-Wcast-qual]
  420 |     const half2 x_reg = *((half2 *) &(x[ib + iqs]));
      |                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:682:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  682 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:524:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_0>' requested here
  524 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:533:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_1>' requested here
  533 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:542:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_0>' requested here
  542 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:551:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_1>' requested here
  551 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:560:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q8_0>' requested here
  560 |     dequantize_mul_mat_vec<GGML_TYPE_Q8_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:611:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_F16>' requested here
  611 |     dequantize_mul_mat_vec<GGML_TYPE_F16>
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
15 warnings generated when compiling for gfx1035.
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:420:37: warning: cast from 'const __half *' to '__half2 *' drops const qualifier [-Wcast-qual]
  420 |     const half2 x_reg = *((half2 *) &(x[ib + iqs]));
      |                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:682:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  682 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:524:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_0>' requested here
  524 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:533:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_1>' requested here
  533 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:542:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_0>' requested here
  542 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:551:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_1>' requested here
  551 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:560:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q8_0>' requested here
  560 |     dequantize_mul_mat_vec<GGML_TYPE_Q8_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:611:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_F16>' requested here
  611 |     dequantize_mul_mat_vec<GGML_TYPE_F16>
      |     ^
6 warnings generated when compiling for gfx1151.
15 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:420:37: warning: cast from 'const __half *' to '__half2 *' drops const qualifier [-Wcast-qual]
  420 |     const half2 x_reg = *((half2 *) &(x[ib + iqs]));
      |                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:682:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  682 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:524:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_0>' requested here
  524 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:533:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_1>' requested here
  533 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:542:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_0>' requested here
  542 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:551:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_1>' requested here
  551 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:560:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q8_0>' requested here
  560 |     dequantize_mul_mat_vec<GGML_TYPE_Q8_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:611:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_F16>' requested here
  611 |     dequantize_mul_mat_vec<GGML_TYPE_F16>
      |     ^
6 warnings generated when compiling for gfx900.
15 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:420:37: warning: cast from 'const __half *' to '__half2 *' drops const qualifier [-Wcast-qual]
  420 |     const half2 x_reg = *((half2 *) &(x[ib + iqs]));
      |                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:682:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  682 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:524:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_0>' requested here
  524 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:533:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_1>' requested here
  533 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:542:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_0>' requested here
  542 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:551:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_1>' requested here
  551 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:560:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q8_0>' requested here
  560 |     dequantize_mul_mat_vec<GGML_TYPE_Q8_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:611:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_F16>' requested here
  611 |     dequantize_mul_mat_vec<GGML_TYPE_F16>
      |     ^
15 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:420:37: warning: cast from 'const __half *' to '__half2 *' drops const qualifier [-Wcast-qual]
  420 |     const half2 x_reg = *((half2 *) &(x[ib + iqs]));
      |                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:682:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  682 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:524:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_0>' requested here
  524 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:533:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_1>' requested here
  533 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:542:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_0>' requested here
  542 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:551:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_1>' requested here
  551 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:560:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q8_0>' requested here
  560 |     dequantize_mul_mat_vec<GGML_TYPE_Q8_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:611:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_F16>' requested here
  611 |     dequantize_mul_mat_vec<GGML_TYPE_F16>
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
15 warnings generated when compiling for gfx1103.
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:420:37: warning: cast from 'const __half *' to '__half2 *' drops const qualifier [-Wcast-qual]
  420 |     const half2 x_reg = *((half2 *) &(x[ib + iqs]));
      |                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:682:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  682 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:524:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_0>' requested here
  524 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:533:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_1>' requested here
  533 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:542:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_0>' requested here
  542 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:551:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_1>' requested here
  551 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:560:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q8_0>' requested here
  560 |     dequantize_mul_mat_vec<GGML_TYPE_Q8_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:611:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_F16>' requested here
  611 |     dequantize_mul_mat_vec<GGML_TYPE_F16>
      |     ^
6 warnings generated when compiling for gfx90a.
15 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:420:37: warning: cast from 'const __half *' to '__half2 *' drops const qualifier [-Wcast-qual]
  420 |     const half2 x_reg = *((half2 *) &(x[ib + iqs]));
      |                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:682:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  682 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:524:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_0>' requested here
  524 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:533:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_1>' requested here
  533 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:542:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_0>' requested here
  542 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:551:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_1>' requested here
  551 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:560:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q8_0>' requested here
  560 |     dequantize_mul_mat_vec<GGML_TYPE_Q8_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:611:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_F16>' requested here
  611 |     dequantize_mul_mat_vec<GGML_TYPE_F16>
      |     ^
15 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 28%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f16.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f16.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f16.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f16.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:420:37: warning: cast from 'const __half *' to '__half2 *' drops const qualifier [-Wcast-qual]
  420 |     const half2 x_reg = *((half2 *) &(x[ib + iqs]));
      |                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:682:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  682 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:524:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_0>' requested here
  524 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:533:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_1>' requested here
  533 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:542:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_0>' requested here
  542 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:551:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_1>' requested here
  551 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:560:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q8_0>' requested here
  560 |     dequantize_mul_mat_vec<GGML_TYPE_Q8_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:611:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_F16>' requested here
  611 |     dequantize_mul_mat_vec<GGML_TYPE_F16>
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
15 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:420:37: warning: cast from 'const __half *' to '__half2 *' drops const qualifier [-Wcast-qual]
  420 |     const half2 x_reg = *((half2 *) &(x[ib + iqs]));
      |                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:682:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  682 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:524:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_0>' requested here
  524 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:533:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_1>' requested here
  533 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:542:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_0>' requested here
  542 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:551:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_1>' requested here
  551 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:560:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q8_0>' requested here
  560 |     dequantize_mul_mat_vec<GGML_TYPE_Q8_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:611:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_F16>' requested here
  611 |     dequantize_mul_mat_vec<GGML_TYPE_F16>
      |     ^
15 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:420:37: warning: cast from 'const __half *' to '__half2 *' drops const qualifier [-Wcast-qual]
  420 |     const half2 x_reg = *((half2 *) &(x[ib + iqs]));
      |                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:682:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  682 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:524:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_0>' requested here
  524 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:533:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_1>' requested here
  533 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:542:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_0>' requested here
  542 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:551:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_1>' requested here
  551 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:560:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q8_0>' requested here
  560 |     dequantize_mul_mat_vec<GGML_TYPE_Q8_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:611:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_F16>' requested here
  611 |     dequantize_mul_mat_vec<GGML_TYPE_F16>
      |     ^
15 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:420:37: warning: cast from 'const __half *' to '__half2 *' drops const qualifier [-Wcast-qual]
  420 |     const half2 x_reg = *((half2 *) &(x[ib + iqs]));
      |                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:682:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  682 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:524:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_0>' requested here
  524 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:533:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_1>' requested here
  533 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:542:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_0>' requested here
  542 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:551:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_1>' requested here
  551 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:560:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q8_0>' requested here
  560 |     dequantize_mul_mat_vec<GGML_TYPE_Q8_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:611:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_F16>' requested here
  611 |     dequantize_mul_mat_vec<GGML_TYPE_F16>
      |     ^
15 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:420:37: warning: cast from 'const __half *' to '__half2 *' drops const qualifier [-Wcast-qual]
  420 |     const half2 x_reg = *((half2 *) &(x[ib + iqs]));
      |                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:682:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  682 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:524:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_0>' requested here
  524 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:533:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_1>' requested here
  533 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:542:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_0>' requested here
  542 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:551:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_1>' requested here
  551 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:560:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q8_0>' requested here
  560 |     dequantize_mul_mat_vec<GGML_TYPE_Q8_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:611:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_F16>' requested here
  611 |     dequantize_mul_mat_vec<GGML_TYPE_F16>
      |     ^
30 warnings generated when compiling for gfx1010.
15 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:420:37: warning: cast from 'const __half *' to '__half2 *' drops const qualifier [-Wcast-qual]
  420 |     const half2 x_reg = *((half2 *) &(x[ib + iqs]));
      |                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:682:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  682 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:524:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_0>' requested here
  524 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:533:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q4_1>' requested here
  533 |     dequantize_mul_mat_vec<GGML_TYPE_Q4_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:542:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_0>' requested here
  542 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:551:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q5_1>' requested here
  551 |     dequantize_mul_mat_vec<GGML_TYPE_Q5_1>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:560:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_Q8_0>' requested here
  560 |     dequantize_mul_mat_vec<GGML_TYPE_Q8_0>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:494:53: warning: cast from 'const float *' to 'HIP_vector_type<float, 2> *' drops const qualifier [-Wcast-qual]
  494 |                 const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
      |                                                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/dmmv.cu:611:5: note: in instantiation of function template specialization 'dequantize_mul_mat_vec<GGML_TYPE_F16>' requested here
  611 |     dequantize_mul_mat_vec<GGML_TYPE_F16>
      |     ^
15 warnings generated when compiling for host.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
[ 29%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f32.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f32.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f32.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f32.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
30 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1030.
30 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1031.
30 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1031.
30 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
30 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
30 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx906.
30 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
30 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for host.
[ 30%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
30 warnings generated when compiling for gfx942.
80 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
80 warnings generated when compiling for gfx1012.
30 warnings generated when compiling for host.
[ 31%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/getrows.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/getrows.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/getrows.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/getrows.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:175:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  175 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1010.
80 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:175:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  175 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
7 warnings generated when compiling for gfx1012.
80 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:175:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  175 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
80 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:175:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  175 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
7 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
80 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:175:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  175 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
80 warnings generated when compiling for gfx1101.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:175:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  175 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
80 warnings generated when compiling for gfx1102.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:175:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  175 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:175:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  175 |             break;
      |             ^~~~~
80 warnings generated when compiling for gfx1103.
7 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:175:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  175 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
7 warnings generated when compiling for gfx1103.
80 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:175:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  175 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
80 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:175:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  175 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h7 warnings generated when compiling for gfx900.
:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
80 warnings generated when compiling for gfx906.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:175:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  175 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
80 warnings generated when compiling for gfx908.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:175:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  175 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:175:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  175 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx90a.
80 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:175:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  175 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
7 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
80 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:175:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  175 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
7 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
80 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/getrows.cu:175:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  175 |             break;
      |             ^~~~~
7 warnings generated when compiling for host.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
[ 32%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/ggml-cuda.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/ggml-cuda.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/ggml-cuda.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/ggml-cuda.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
80 warnings generated when compiling for host.
[ 33%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/im2col.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/im2col.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/im2col.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/im2col.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml-cuda.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:584:12: warning: 'backend' is deprecated: use the buffer type to find the storage location of the tensor [-Wdeprecated-declarations]
  584 |     struct ggml_tensor {
      |            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2019:28: note: in implicit copy constructor for 'ggml_tensor' first required here
 2019 |     ggml_tensor src0_row = *src0;
      |                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:587:9: note: 'backend' has been explicitly marked deprecated here
  587 |         GGML_DEPRECATED(enum ggml_backend_type backend, "use the buffer type to find the storage location of the tensor");
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:192:61: note: expanded from macro 'GGML_DEPRECATED'
  192 | #    define GGML_DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
      |                                                             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:67: warning: unused parameter 'size' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3130:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3130 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3126:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3126 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3117:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3117 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3110:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3110 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3105:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3105 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3100:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3100 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3095:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3095 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3053:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3053 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3036:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3036 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2975:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2975 |             break;
      |             ^~~~~
6 warnings generated when compiling for gfx1010.
27 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml-cuda.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:584:12: warning: 'backend' is deprecated: use the buffer type to find the storage location of the tensor [-Wdeprecated-declarations]
  584 |     struct ggml_tensor {
      |            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2019:28: note: in implicit copy constructor for 'ggml_tensor' first required here
 2019 |     ggml_tensor src0_row = *src0;
      |                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:587:9: note: 'backend' has been explicitly marked deprecated here
  587 |         GGML_DEPRECATED(enum ggml_backend_type backend, "use the buffer type to find the storage location of the tensor");
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:192:61: note: expanded from macro 'GGML_DEPRECATED'
  192 | #    define GGML_DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
      |                                                             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:67: warning: unused parameter 'size' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3130:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3130 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3126:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3126 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3117:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3117 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3110:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3110 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3105:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3105 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3100:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3100 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3095:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3095 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3053:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3053 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3036:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3036 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2975:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2975 |             break;
      |             ^~~~~
27 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml-cuda.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:584:12: warning: 'backend' is deprecated: use the buffer type to find the storage location of the tensor [-Wdeprecated-declarations]
  584 |     struct ggml_tensor {
      |            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2019:28: note: in implicit copy constructor for 'ggml_tensor' first required here
 2019 |     ggml_tensor src0_row = *src0;
      |                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:587:9: note: 'backend' has been explicitly marked deprecated here
  587 |         GGML_DEPRECATED(enum ggml_backend_type backend, "use the buffer type to find the storage location of the tensor");
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:192:61: note: expanded from macro 'GGML_DEPRECATED'
  192 | #    define GGML_DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
      |                                                             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:67: warning: unused parameter 'size' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3130:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3130 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3126:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3126 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3117:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3117 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3110:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3110 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3105:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3105 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3100:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3100 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3095:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3095 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3053:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3053 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3036:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3036 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2975:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2975 |             break;
      |             ^~~~~
27 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml-cuda.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:584:12: warning: 'backend' is deprecated: use the buffer type to find the storage location of the tensor [-Wdeprecated-declarations]
  584 |     struct ggml_tensor {
      |            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2019:28: note: in implicit copy constructor for 'ggml_tensor' first required here
 2019 |     ggml_tensor src0_row = *src0;
      |                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:587:9: note: 'backend' has been explicitly marked deprecated here
  587 |         GGML_DEPRECATED(enum ggml_backend_type backend, "use the buffer type to find the storage location of the tensor");
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:192:61: note: expanded from macro 'GGML_DEPRECATED'
  192 | #    define GGML_DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
      |                                                             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:67: warning: unused parameter 'size' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3130:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3130 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3126:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3126 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3117:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3117 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3110:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3110 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3105:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3105 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3100:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3100 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3095:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3095 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3053:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3053 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3036:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3036 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2975:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2975 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
27 warnings generated when compiling for gfx1031.
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml-cuda.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:584:12: warning: 'backend' is deprecated: use the buffer type to find the storage location of the tensor [-Wdeprecated-declarations]
  584 |     struct ggml_tensor {
      |            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2019:28: note: in implicit copy constructor for 'ggml_tensor' first required here
 2019 |     ggml_tensor src0_row = *src0;
      |                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:587:9: note: 'backend' has been explicitly marked deprecated here
  587 |         GGML_DEPRECATED(enum ggml_backend_type backend, "use the buffer type to find the storage location of the tensor");
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:192:61: note: expanded from macro 'GGML_DEPRECATED'
  192 | #    define GGML_DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
      |                                                             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:67: warning: unused parameter 'size' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3130:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3130 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3126:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3126 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3117:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3117 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3110:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3110 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3105:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3105 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3100:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3100 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3095:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3095 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3053:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3053 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3036:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3036 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2975:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2975 |             break;
      |             ^~~~~
27 warnings generated when compiling for gfx1035.
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml-cuda.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:584:12: warning: 'backend' is deprecated: use the buffer type to find the storage location of the tensor [-Wdeprecated-declarations]
  584 |     struct ggml_tensor {
      |            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2019:28: note: in implicit copy constructor for 'ggml_tensor' first required here
 2019 |     ggml_tensor src0_row = *src0;
      |                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:587:9: note: 'backend' has been explicitly marked deprecated here
  587 |         GGML_DEPRECATED(enum ggml_backend_type backend, "use the buffer type to find the storage location of the tensor");
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:192:61: note: expanded from macro 'GGML_DEPRECATED'
  192 | #    define GGML_DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
      |                                                             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:67: warning: unused parameter 'size' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3130:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3130 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3126:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3126 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3117:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3117 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3110:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3110 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3105:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3105 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3100:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3100 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3095:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3095 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3053:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3053 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3036:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3036 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2975:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2975 |             break;
      |             ^~~~~
27 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml-cuda.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:584:12: warning: 'backend' is deprecated: use the buffer type to find the storage location of the tensor [-Wdeprecated-declarations]
  584 |     struct ggml_tensor {
      |            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2019:28: note: in implicit copy constructor for 'ggml_tensor' first required here
 2019 |     ggml_tensor src0_row = *src0;
      |                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:587:9: note: 'backend' has been explicitly marked deprecated here
  587 |         GGML_DEPRECATED(enum ggml_backend_type backend, "use the buffer type to find the storage location of the tensor");
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:192:61: note: expanded from macro 'GGML_DEPRECATED'
  192 | #    define GGML_DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
      |                                                             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:67: warning: unused parameter 'size' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3130:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3130 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3126:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3126 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3117:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3117 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3110:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3110 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3105:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3105 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3100:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3100 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3095:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3095 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3053:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3053 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3036:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3036 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2975:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2975 |             break;
      |             ^~~~~
27 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml-cuda.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:584:12: warning: 'backend' is deprecated: use the buffer type to find the storage location of the tensor [-Wdeprecated-declarations]
  584 |     struct ggml_tensor {
      |            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2019:28: note: in implicit copy constructor for 'ggml_tensor' first required here
 2019 |     ggml_tensor src0_row = *src0;
      |                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:587:9: note: 'backend' has been explicitly marked deprecated here
  587 |         GGML_DEPRECATED(enum ggml_backend_type backend, "use the buffer type to find the storage location of the tensor");
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:192:61: note: expanded from macro 'GGML_DEPRECATED'
  192 | #    define GGML_DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
      |                                                             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:67: warning: unused parameter 'size' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3130:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3130 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3126:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3126 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3117:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3117 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3110:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3110 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3105:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3105 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3100:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3100 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3095:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3095 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3053:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3053 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3036:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3036 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2975:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2975 |             break;
      |             ^~~~~
27 warnings generated when compiling for gfx1102.
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml-cuda.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:584:12: warning: 'backend' is deprecated: use the buffer type to find the storage location of the tensor [-Wdeprecated-declarations]
  584 |     struct ggml_tensor {
      |            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2019:28: note: in implicit copy constructor for 'ggml_tensor' first required here
 2019 |     ggml_tensor src0_row = *src0;
      |                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:587:9: note: 'backend' has been explicitly marked deprecated here
  587 |         GGML_DEPRECATED(enum ggml_backend_type backend, "use the buffer type to find the storage location of the tensor");
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:192:61: note: expanded from macro 'GGML_DEPRECATED'
  192 | #    define GGML_DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
      |                                                             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:67: warning: unused parameter 'size' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3130:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3130 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3126:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3126 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3117:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3117 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3110:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3110 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3105:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3105 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3100:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3100 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3095:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3095 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3053:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3053 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3036:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3036 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2975:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2975 |             break;
      |             ^~~~~
6 warnings generated when compiling for gfx900.
27 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml-cuda.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:584:12: warning: 'backend' is deprecated: use the buffer type to find the storage location of the tensor [-Wdeprecated-declarations]
  584 |     struct ggml_tensor {
      |            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2019:28: note: in implicit copy constructor for 'ggml_tensor' first required here
 2019 |     ggml_tensor src0_row = *src0;
      |                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:587:9: note: 'backend' has been explicitly marked deprecated here
  587 |         GGML_DEPRECATED(enum ggml_backend_type backend, "use the buffer type to find the storage location of the tensor");
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:192:61: note: expanded from macro 'GGML_DEPRECATED'
  192 | #    define GGML_DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
      |                                                             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:67: warning: unused parameter 'size' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3130:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3130 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3126:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3126 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3117:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3117 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3110:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3110 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3105:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3105 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3100:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3100 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3095:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3095 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3053:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3053 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3036:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3036 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2975:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2975 |             break;
      |             ^~~~~
27 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml-cuda.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:584:12: warning: 'backend' is deprecated: use the buffer type to find the storage location of the tensor [-Wdeprecated-declarations]
  584 |     struct ggml_tensor {
      |            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2019:28: note: in implicit copy constructor for 'ggml_tensor' first required here
 2019 |     ggml_tensor src0_row = *src0;
      |                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:587:9: note: 'backend' has been explicitly marked deprecated here
  587 |         GGML_DEPRECATED(enum ggml_backend_type backend, "use the buffer type to find the storage location of the tensor");
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:192:61: note: expanded from macro 'GGML_DEPRECATED'
  192 | #    define GGML_DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
      |                                                             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:67: warning: unused parameter 'size' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3130:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3130 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3126:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3126 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3117:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3117 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3110:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3110 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3105:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3105 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3100:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3100 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3095:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3095 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3053:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3053 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3036:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3036 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2975:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2975 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
27 warnings generated when compiling for gfx900.
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml-cuda.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:584:12: warning: 'backend' is deprecated: use the buffer type to find the storage location of the tensor [-Wdeprecated-declarations]
  584 |     struct ggml_tensor {
      |            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2019:28: note: in implicit copy constructor for 'ggml_tensor' first required here
 2019 |     ggml_tensor src0_row = *src0;
      |                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:587:9: note: 'backend' has been explicitly marked deprecated here
  587 |         GGML_DEPRECATED(enum ggml_backend_type backend, "use the buffer type to find the storage location of the tensor");
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:192:61: note: expanded from macro 'GGML_DEPRECATED'
  192 | #    define GGML_DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
      |                                                             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:67: warning: unused parameter 'size' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3130:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3130 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3126:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3126 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3117:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3117 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3110:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3110 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3105:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3105 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3100:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3100 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3095:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3095 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3053:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3053 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3036:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3036 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2975:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2975 |             break;
      |             ^~~~~
27 warnings generated when compiling for gfx906.
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml-cuda.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:584:12: warning: 'backend' is deprecated: use the buffer type to find the storage location of the tensor [-Wdeprecated-declarations]
  584 |     struct ggml_tensor {
      |            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2019:28: note: in implicit copy constructor for 'ggml_tensor' first required here
 2019 |     ggml_tensor src0_row = *src0;
      |                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:587:9: note: 'backend' has been explicitly marked deprecated here
  587 |         GGML_DEPRECATED(enum ggml_backend_type backend, "use the buffer type to find the storage location of the tensor");
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:192:61: note: expanded from macro 'GGML_DEPRECATED'
  192 | #    define GGML_DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
      |                                                             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:67: warning: unused parameter 'size' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3130:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3130 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3126:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3126 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3117:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3117 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3110:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3110 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3105:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3105 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3100:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3100 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3095:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3095 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3053:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3053 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3036:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3036 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2975:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2975 |             break;
      |             ^~~~~
27 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 34%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmq.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmq.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmq.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmq.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml-cuda.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:584:12: warning: 'backend' is deprecated: use the buffer type to find the storage location of the tensor [-Wdeprecated-declarations]
  584 |     struct ggml_tensor {
      |            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2019:28: note: in implicit copy constructor for 'ggml_tensor' first required here
 2019 |     ggml_tensor src0_row = *src0;
      |                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:587:9: note: 'backend' has been explicitly marked deprecated here
  587 |         GGML_DEPRECATED(enum ggml_backend_type backend, "use the buffer type to find the storage location of the tensor");
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:192:61: note: expanded from macro 'GGML_DEPRECATED'
  192 | #    define GGML_DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
      |                                                             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:67: warning: unused parameter 'size' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3130:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3130 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3126:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3126 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3117:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3117 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3110:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3110 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3105:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3105 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3100:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3100 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3095:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3095 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3053:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3053 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3036:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3036 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2975:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2975 |             break;
      |             ^~~~~
27 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml-cuda.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:584:12: warning: 'backend' is deprecated: use the buffer type to find the storage location of the tensor [-Wdeprecated-declarations]
  584 |     struct ggml_tensor {
      |            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2019:28: note: in implicit copy constructor for 'ggml_tensor' first required here
 2019 |     ggml_tensor src0_row = *src0;
      |                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:587:9: note: 'backend' has been explicitly marked deprecated here
  587 |         GGML_DEPRECATED(enum ggml_backend_type backend, "use the buffer type to find the storage location of the tensor");
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:192:61: note: expanded from macro 'GGML_DEPRECATED'
  192 | #    define GGML_DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
      |                                                             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:67: warning: unused parameter 'size' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3130:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3130 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3126:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3126 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3117:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3117 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3110:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3110 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3105:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3105 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3100:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3100 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3095:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3095 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3053:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3053 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3036:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3036 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2975:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2975 |             break;
      |             ^~~~~
27 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml-cuda.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:584:12: warning: 'backend' is deprecated: use the buffer type to find the storage location of the tensor [-Wdeprecated-declarations]
  584 |     struct ggml_tensor {
      |            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2019:28: note: in implicit copy constructor for 'ggml_tensor' first required here
 2019 |     ggml_tensor src0_row = *src0;
      |                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:587:9: note: 'backend' has been explicitly marked deprecated here
  587 |         GGML_DEPRECATED(enum ggml_backend_type backend, "use the buffer type to find the storage location of the tensor");
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:192:61: note: expanded from macro 'GGML_DEPRECATED'
  192 | #    define GGML_DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
      |                                                             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:67: warning: unused parameter 'size' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3130:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3130 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3126:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3126 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3117:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3117 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3110:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3110 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3105:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3105 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3100:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3100 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3095:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3095 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3053:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3053 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3036:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3036 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2975:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2975 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1030.
27 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:23:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml-cuda.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:584:12: warning: 'backend' is deprecated: use the buffer type to find the storage location of the tensor [-Wdeprecated-declarations]
  584 |     struct ggml_tensor {
      |            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2019:28: note: in implicit copy constructor for 'ggml_tensor' first required here
 2019 |     ggml_tensor src0_row = *src0;
      |                            ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:587:9: note: 'backend' has been explicitly marked deprecated here
  587 |         GGML_DEPRECATED(enum ggml_backend_type backend, "use the buffer type to find the storage location of the tensor");
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include/ggml.h:192:61: note: expanded from macro 'GGML_DEPRECATED'
  192 | #    define GGML_DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
      |                                                             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2834:67: warning: unused parameter 'size' [-Wunused-parameter]
 2834 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3130:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3130 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3126:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3126 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3117:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3117 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3110:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3110 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3105:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3105 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3100:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3100 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3095:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3095 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3053:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3053 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:3036:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3036 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/ggml-cuda.cu:2975:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2975 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
27 warnings generated when compiling for host.
[ 35%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmvq.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmvq.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmvq.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmvq.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:417:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  417 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for host.
[ 36%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/norm.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/norm.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/norm.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/norm.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 37%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/opt-step-adamw.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/opt-step-adamw.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/opt-step-adamw.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/opt-step-adamw.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
293 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:417:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  417 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^

  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<</builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.hb:l305o:c9k:_ nwarning: uanonymous types declared in an anonymous union are an extension [-Wnested-anon-types]m
s,   b305l | o c k _ d i m s ,s t0r,u cstt r{e
a      m|         ^
>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 38%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/out-prod.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/out-prod.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/out-prod.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/out-prod.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 39%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pad.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pad.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/pad.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/pad.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1103.
293 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1151.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:417:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  417 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for host.
[ 40%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pool2d.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pool2d.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/pool2d.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/pool2d.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
161 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:417:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  417 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_T/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.hY:P196E:_9Q4_0>: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
(vx, vy, ds/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
t, n/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
cols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h>:>(305v:x9,:  vwarning: yanonymous types declared in an anonymous union are an extension [-Wnested-anon-types],
 dst  ,305  | n c o l s _ x ,  sntrrouwcst_ x{,
       n| r        ^o
ws_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
161 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:417:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  417 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 41%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/quantize.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/quantize.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/quantize.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/quantize.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1102.
161 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:417:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  417 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.hn:r196o:w9s:_ dwarning: sanonymous types declared in an anonymous union are an extension [-Wnested-anon-types]t
, st  r196e | a m ) ; 
        |      ^ 
struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h<:G261G:M9L:_ Twarning: Yanonymous types declared in an anonymous union are an extension [-Wnested-anon-types]P
E_IQ  4261_ | N L > ( v x ,   vsyt,r udcstt ,{ 
n      c| o        ^l
s_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for host.
[ 42%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/rope.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/rope.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/rope.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/rope.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
161 warnings generated when compiling for gfx1100.
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:417:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  417 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
 |     /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         structf l{o
a      t|          ^t
mp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  178 |             mul_mat_vec_q</builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
161 warnings generated when compiling for gfx1101.
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:417:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  417 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 43%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/scale.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/scale.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/scale.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/scale.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
161 warnings generated when compiling for gfx1102.
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:417:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  417 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
161 warnings generated when compiling for gfx1103.
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:417:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  417 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
6 warnings generated when compiling for host.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream[ 44%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/softmax.cu.o
);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/softmax.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/softmax.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/softmax.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
18 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
18 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
18 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
18 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
18 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
18 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
161 warnings generated when compiling for gfx1151.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
18 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:417:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  417 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(In file included from vx/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu,: 1v:
yIn file included from ,/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh :d20s:
t,/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h n:c154o:l9s:_ xwarning: ,anonymous types declared in an anonymous union are an extension [-Wnested-anon-types] 
nrows_x, nrows_y  ,154  | n c o l s _ y ,  sntrrouwcst_ d{s
t      ,|          ^s
tream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
18 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
18 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
18 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
18 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
18 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:37:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   37 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:73:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
   73 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:106:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
  106 |     for (int col0 = 0; col0 < ncols; col0 += block_size) {
      |     ^
18 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 45%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/sum.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/sum.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/sum.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/sum.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sum.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sum.cu:12:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sum.cu:12:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sum.cu:12:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sum.cu:12:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sum.cu:12:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sum.cu:12:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sum.cu:12:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sum.cu:12:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sum.cu:12:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sum.cu:12:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sum.cu:12:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sum.cu:12:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sum.cu:12:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sum.cu:12:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sum.cu:12:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sum.cu:12:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sum.cu:12:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 46%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/sumrows.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/sumrows.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/sumrows.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/sumrows.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
293 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:417:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  417 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 47%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/tsembd.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/tsembd.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/tsembd.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/tsembd.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 48%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/unary.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/unary.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/unary.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/unary.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
293 warnings generated when compiling for gfx906.
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:417:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  417 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 49%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/upscale.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/upscale.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/upscale.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/upscale.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for host.
[ 50%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/wkv6.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/wkv6.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/wkv6.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/wkv6.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/wkv6.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
293 warnings generated when compiling for gfx908.
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:417:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  417 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 51%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
293 warnings generated when compiling for gfx90a.
64 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:417:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  417 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu::23126:: 98warning: :comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare] 
warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  318   | 126  |       f o r   ( iinft  (lt h=r e0a;d Ild x<. xs i<z eroofw(si_npte)r;_ c+u+dla)_ b{l
o      c| k                    ~ ^ ~~~~~~~~~~~ 
&& (r/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuho:w325s:_23p:e rwarning: _comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]c
uda_b  l325o | c k   = =f o1r  |(|i nrto wl0  =+  1t;h rle a<d Isdixz.exo f<( innrto)w;s _+d+slt)) ){ 
{      
|                           ~ ^ ~~~~~~~~~~~| 
                                                                              ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cucast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~

  554/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
 |      /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (   *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(cons/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
t int/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
 D) /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx,{
      |                                               ^
 vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restric/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
t__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const cha,r  *n r_o_wrse_syt,r ncioclts___ yK,, 
n      r| o                                  ^w
s/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh_d:s16:t35,:  swarning: tunused parameter 'V' [-Wunused-parameter]rea
m   )16; | 
           |      ^
 /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict_:_ 126:98d:s t,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19:warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |    warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
  mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (thrIn file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
eadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuIn file included from :/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh126::298:
:/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh :warning: 554comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]:
24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
    554126 |  |                 *i(f( u(itnhtr3e2a_dtI d*x). x& K<Q _rmoawxs__spcearl_ec)u d&a=_ bfltozc_km a&s&k ;(
r      o| w                       ^s
_per_/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuhc:u704d:a5_:b lnote: oin instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested herec
k ==   7041 |   | |   rfolwa0s h+_ atthtrne_acdoImdbxi.nxe _<r ensruolwtss_<dDs,t )p)a r{a
l      l| e                                                                              ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~l
_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126  :49198 | :   warning:  comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare] 
    launch_  f126a | t t n < D ,   p airfa l(ltehlr_ebaldoIcdkxs.>x( c<t xr,o wdss_tp,e rf_actutdna__kbelroncekl ,& &n w(arropwss,_ pceorl_sc_updear__bblloocckk ,= =t r1u e|,|  trrouwe0) ;+
       t| h        ^r
eadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuIn file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554::12624::98 :warning:  cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]warning: 
comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  554 |          126  | * ( ( u i n t 3 2i_ft  (*t)h r&eKaQd_Imdaxx._xs c<a lreo)w s&_=p efrt_zc_umdaas_kb;l
o      c| k                       ^ 
&& (r/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuho:w704s:_5p:e rnote: _in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested herec
uda_  b704l | o c k   =f=l a1s h|_|a trtonw_0c o+m btihnree_ardeIsduxl.txs <<D ,n rpoawrsa_ldlsetl)_)b l{o
c      k| s                                                                              ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~>

      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
64 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for host.
[ 52%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
293 warnings generated when compiling for gfx90a.
61 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:417:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  417 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h,: n261r:o9w:s _warning: xanonymous types declared in an anonymous union are an extension [-Wnested-anon-types],
 nro  w261s | _ y ,   n c o l ss_tyr,u cntr o{w
s      _| d        ^s
t, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~

  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nco/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuhls:_554y:,24 :n rwarning: ocast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]w
s_dst,   s554t | r e a m ) ; 
        | *    ^(
(uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restr  i126ct | _ _   d s t _ m eitfa ,(
t      h| r                                  ^e
adI/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuhd:20x:.21x:  <warning:  unused parameter 'scale' [-Wunused-parameter]r
ows   _20p | e r _ c u d a _ bclooncskt  &f&l o(arto wssc_apleer,_
c      u|                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |     d a _ b lcooncskt  =f=l o1a t |m|a xr_obwi0a s+, 
t      h| r                    ^e
adI/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuhd:22x:.21x:  <warning: unused parameter 'm0' [-Wunused-parameter] 
nrow   22s | _d s t ) )   { 
       c| o                                                                              ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~n
st float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |     /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu  : 80 :c48o:n swarning: tsuggest braces around initialization of subobject [-Wmissing-braces] 
int nb31,   
80       | |                    ^
   /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuhf:l36o:a19t:  twarning: munused parameter 'nb01' [-Wunused-parameter]p
[nc   o36l | s _ y ] [ r o w sc_opnesrt_ ciundta _nbbl0o1,c
k      ]|                    ^=
 {/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh0.:037:f19}:; 
warning:       unused parameter 'nb02' [-Wunused-parameter]
|                                                ^~~~
         37|  |                                                {   } 
       /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuco:n184:st13: i note: nin instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested heret 
nb02,
      |                     ^184
 |  /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh : 38 : 19 :   warning:  unused parameter 'nb03' [-Wunused-parameter] 
  m   u38 | l _ m a t _v  e cc_oqn<stty pien,t  4n>b<0<3<,b
l      o| c                  ^k
_nu/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:m39s,: 19b:l owarning: cunused parameter 'nb11' [-Wunused-parameter]k
_di   m39 | s ,   0 ,   s t rceoanmst> >in>t( vnxb,1 1,v
y      ,|                    ^d
st/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh, :n40c:o19l:s warning: _unused parameter 'nb12' [-Wunused-parameter]x,
 nr   40ow |  s_  x ,   n r ocwosn_syt,  innrto wnbs1_2d,s
t      )| ;                  ^

      | /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh            ^:41
:19:/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu :warning: 313:unused parameter 'nb13' [-Wunused-parameter]5
: note:    in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here41 | 
          313c | o n s t  miunlt_ mnabt1_3v,e
c      _| q                  ^_
cud/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:a42<:G19GM:L _warning: unused parameter 'nb21' [-Wunused-parameter]T
YPE   _42IQ | 1  _ M> (  v x ,c ovnys,t  disntt,  nnbc2o1l,s
_      x| ,                  ^ 
nr/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuho:w43s:_19x:,  warning: nunused parameter 'nb22' [-Wunused-parameter]r
ows   _43y | ,   n c o l s _ yc,o nnsrto wisn_td sntb,2 2s,t
r      e| a                  ^m
);/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh
:      44| :    ^19
: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^

   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for host.
[ 53%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for host.
[ 54%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1031.
293 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
64 warnings generated when compiling for gfx1035.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:417:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  417 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:208:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  208 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:215:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  215 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:222:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  222 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:229:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  229 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:236:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  236 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:243:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  243 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:250:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  250 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:257:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  257 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:264:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  264 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:271:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  271 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:278:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  278 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:285:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  285 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:292:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  292 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:299:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  299 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:306:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  306 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:313:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  313 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:320:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  320 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:327:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  327 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:175:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  175 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:178:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  178 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:181:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  181 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:184:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  184 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:187:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  187 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:190:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  190 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:193:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  193 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:196:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  196 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:334:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  334 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
293 warnings generated when compiling for host.
[ 55%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu
64 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx1010.
64 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx1012.
64 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx1030.
64 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
58 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
58 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
64 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
58 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
64 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
58 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
64 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
58 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
64 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
64 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx942.
58 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for host.
[ 56%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq1_s.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq1_s.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq1_s.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq1_s.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu
58 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.xIn file included from  |/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu|: j3t:
 In file included from !/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh=: 2b:
lo/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuhc:kI554d:x24.:y )warning:  cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]{

      |             ~~ ^  ~~~~~~~~~~
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
58 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for host.
[ 56%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_s.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_s.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_s.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_s.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 57%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 58%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 59%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_s.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_s.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_s.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_s.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 60%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 61%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 62%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 63%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q2_k.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q2_k.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q2_k.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q2_k.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 64%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q3_k.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q3_k.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q3_k.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q3_k.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
 1022 |     for (int k01 = 0; k01 < WARP_SIZE; k01 += QR2_K*VDR_Q2_K_Q8_1_MMQ) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
94 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
 1022 |     for (int k01 = 0; k01 < WARP_SIZE; k01 += QR2_K*VDR_Q2_K_Q8_1_MMQ) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
118 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 65%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_0.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_0.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_0.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_0.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 66%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_1.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_1.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_1.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_1.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 782691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
 warnings generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 67%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_k.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_k.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_k.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_k.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
 1022 |     for (int k01 = 0; k01 < WARP_SIZE; k01 += QR2_K*VDR_Q2_K_Q8_1_MMQ) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
94 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 68%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_0.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_0.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_0.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_0.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 69%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_1.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_1.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_1.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_1.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
 1022 |     for (int k01 = 0; k01 < WARP_SIZE; k01 += QR2_K*VDR_Q2_K_Q8_1_MMQ) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
134 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 70%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_k.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_k.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_k.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_k.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 71%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q6_k.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q6_k.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q6_k.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q6_k.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 72%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q8_0.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q8_0.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q8_0.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q8_0.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 73%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
60 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for host.
[ 74%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for host.
[ 75%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
60 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:344:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  344 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:357:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  357 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:360:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  360 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:370:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  370 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:373:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  373 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:385:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  385 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for host.
[ 76%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
60 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:312:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  312 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:322:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  322 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:325:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  325 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:335:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  335 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:338:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  338 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:348:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  348 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:351:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  351 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:363:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  363 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for host.
[ 77%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
78 warnings generated when compiling for gfx1151.
34 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for host.
[ 78%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for host.
[ 79%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:304:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  304 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:331:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  331 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:307:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  307 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:382:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  382 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for host.
[ 80%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
34 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for host.
[ 81%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for host.
[ 82%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DK_QUANTS_PER_ITERATION=2 -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:285:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  285 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:309:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  309 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:288:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  288 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:360:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  360 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for host.
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:154:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  154 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:175:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  175 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:196:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  196 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:261:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  261 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:288:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  288 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-hip/../ggml-common.h:305:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  305 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 83%] Linking CXX shared library libggml-hip.so
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/cmake -E cmake_link_script CMakeFiles/ggml-hip.dir/link.txt --verbose=1
/usr/bin/hipcc -fPIC -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -Wl,-z,relro -Wl,--as-needed  -Wl,-z,pack-relative-relocs -Wl,-z,now -Wl,-z,now   -Wl,--build-id=sha1 -specs=/usr/lib/rpm/redhat/redhat-package-notes -shared -Wl,-soname,libggml-hip.so -o libggml-hip.so "CMakeFiles/ggml-hip.dir/__/ggml-cuda/acc.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/arange.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/argmax.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/argsort.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/binbcast.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/clamp.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/concat.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv-transpose-1d.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/convert.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/count-equal.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/cpy.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/cross-entropy-loss.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/diagmask.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/dmmv.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f16.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f32.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/getrows.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/ggml-cuda.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/im2col.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmq.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmvq.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/norm.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/opt-step-adamw.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/out-prod.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/pad.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/pool2d.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/quantize.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/rope.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/scale.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/softmax.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/sum.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/sumrows.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/tsembd.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/unary.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/upscale.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/wkv6.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq1_s.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_s.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_s.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q2_k.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q3_k.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_0.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_1.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_k.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_0.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_1.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_k.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q6_k.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q8_0.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu.o"  ../libggml-base.so.b4094 /usr/lib64/libhipblas.so.2.2 --hip-link --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1151 /usr/lib64/librocblas.so.4.2 /usr/lib64/libamdhip64.so.6.2.41134
clang++: warning: argument unused during compilation: '-Xarch_host -fstack-protector-strong' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-Xarch_host -fcf-protection' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
[ 83%] Built target ggml-hip
/usr/bin/gmake  -f ggml/src/CMakeFiles/ggml.dir/build.make ggml/src/CMakeFiles/ggml.dir/depend
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build && /usr/bin/cmake -E cmake_depends "Unix Makefiles" /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094 /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src/CMakeFiles/ggml.dir/DependInfo.cmake "--color="
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
/usr/bin/gmake  -f ggml/src/CMakeFiles/ggml.dir/build.make ggml/src/CMakeFiles/ggml.dir/build
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
[ 84%] Building CXX object ggml/src/CMakeFiles/ggml.dir/ggml-backend-reg.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++11 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT ggml/src/CMakeFiles/ggml.dir/ggml-backend-reg.cpp.o -MF CMakeFiles/ggml.dir/ggml-backend-reg.cpp.o.d -o CMakeFiles/ggml.dir/ggml-backend-reg.cpp.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/ggml-backend-reg.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 85%] Linking CXX shared library libggml.so
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src && /usr/bin/cmake -E cmake_link_script CMakeFiles/ggml.dir/link.txt --verbose=1
/usr/bin/hipcc -fPIC -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -Wl,-z,relro -Wl,--as-needed  -Wl,-z,pack-relative-relocs -Wl,-z,now -Wl,-z,now   -Wl,--build-id=sha1 -specs=/usr/lib/rpm/redhat/redhat-package-notes -shared -Wl,-soname,libggml.so.b4094 -o libggml.so.b4094 "CMakeFiles/ggml.dir/ggml-backend-reg.cpp.o"  ggml-cpu/libggml-cpu.so ggml-hip/libggml-hip.so libggml-base.so.b4094
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
clang++: warning: argument unused during compilation: '-Xarch_host -fstack-protector-strong' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-Xarch_host -fcf-protection' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/ggml/src && /usr/bin/cmake -E cmake_symlink_library libggml.so.b4094 libggml.so.b4094 libggml.so
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
[ 85%] Built target ggml
/usr/bin/gmake  -f src/CMakeFiles/llama.dir/build.make src/CMakeFiles/llama.dir/depend
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build && /usr/bin/cmake -E cmake_depends "Unix Makefiles" /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094 /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/src /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/src/CMakeFiles/llama.dir/DependInfo.cmake "--color="
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
/usr/bin/gmake  -f src/CMakeFiles/llama.dir/build.make src/CMakeFiles/llama.dir/build
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
[ 86%] Building CXX object src/CMakeFiles/llama.dir/llama-vocab.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/../include -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -MD -MT src/CMakeFiles/llama.dir/llama-vocab.cpp.o -MF CMakeFiles/llama.dir/llama-vocab.cpp.o.d -o CMakeFiles/llama.dir/llama-vocab.cpp.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-vocab.cpp
[ 87%] Building CXX object src/CMakeFiles/llama.dir/llama.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/../include -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -MD -MT src/CMakeFiles/llama.dir/llama.cpp.o -MF CMakeFiles/llama.dir/llama.cpp.o.d -o CMakeFiles/llama.dir/llama.cpp.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-vocab.cpp:114:13: warning: unused function 'llama_is_unknown_token' [-Wunused-function]
  114 | static bool llama_is_unknown_token(const llama_vocab & vocab, llama_token id) {
      |             ^~~~~~~~~~~~~~~~~~~~~~
1 warning generated when compiling for gfx906.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-vocab.cpp:114:13: warning: unused function 'llama_is_unknown_token' [-Wunused-function]
  114 | static bool llama_is_unknown_token(const llama_vocab & vocab, llama_token id) {
      |             ^~~~~~~~~~~~~~~~~~~~~~
1 warning generated when compiling for host.
[ 88%] Building CXX object src/CMakeFiles/llama.dir/llama-grammar.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/../include -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -MD -MT src/CMakeFiles/llama.dir/llama-grammar.cpp.o -MF CMakeFiles/llama.dir/llama-grammar.cpp.o.d -o CMakeFiles/llama.dir/llama-grammar.cpp.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-grammar.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama.cpp:16701:13: warning: unused function 'llama_set_s_copy' [-Wunused-function]
 16701 | static void llama_set_s_copy(llama_context & lctx) {
       |             ^~~~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-grammar.cpp:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-grammar.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-impl.h:52:13: warning: unused function 'replace_all' [-Wunused-function]
   52 | static void replace_all(std::string & s, const std::string & search, const std::string & replace) {
      |             ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-grammar.cpp:201:13: warning: unused function 'print_rule_binary' [-Wunused-function]
  201 | static void print_rule_binary(FILE * file, const llama_grammar_rule & rule) {
      |             ^~~~~~~~~~~~~~~~~
2 warnings generated when compiling for gfx906.
1 warning generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-grammar.cpp:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-grammar.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-impl.h:52:13: warning: unused function 'replace_all' [-Wunused-function]
   52 | static void replace_all(std::string & s, const std::string & search, const std::string & replace) {
      |             ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-grammar.cpp:201:13: warning: unused function 'print_rule_binary' [-Wunused-function]
  201 | static void print_rule_binary(FILE * file, const llama_grammar_rule & rule) {
      |             ^~~~~~~~~~~~~~~~~
2 warnings generated when compiling for host.
[ 89%] Building CXX object src/CMakeFiles/llama.dir/llama-sampling.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/../include -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -MD -MT src/CMakeFiles/llama.dir/llama-sampling.cpp.o -MF CMakeFiles/llama.dir/llama-sampling.cpp.o.d -o CMakeFiles/llama.dir/llama-sampling.cpp.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-sampling.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-sampling.cpp:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-sampling.h:5:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-grammar.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-impl.h:52:13: warning: unused function 'replace_all' [-Wunused-function]
   52 | static void replace_all(std::string & s, const std::string & search, const std::string & replace) {
      |             ^~~~~~~~~~~
1 warning generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-sampling.cpp:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-sampling.h:5:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-grammar.h:3:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama-impl.h:52:13: warning: unused function 'replace_all' [-Wunused-function]
   52 | static void replace_all(std::string & s, const std::string & search, const std::string & replace) {
      |             ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/llama.cpp:16701:13: warning: unused function 'llama_set_s_copy' [-Wunused-function]
 16701 | static void llama_set_s_copy(llama_context & lctx) {
       |             ^~~~~~~~~~~~~~~~
1 warning generated when compiling for host.
[ 90%] Building CXX object src/CMakeFiles/llama.dir/unicode.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/../include -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -MD -MT src/CMakeFiles/llama.dir/unicode.cpp.o -MF CMakeFiles/llama.dir/unicode.cpp.o.d -o CMakeFiles/llama.dir/unicode.cpp.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/unicode.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/unicode.cpp:29:20: warning: unused function 'unicode_cpts_to_utf8' [-Wunused-function]
   29 | static std::string unicode_cpts_to_utf8(const std::vector<uint32_t> & cps) {
      |                    ^~~~~~~~~~~~~~~~~~~~
1 warning generated when compiling for gfx906.
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/unicode.cpp:29:20: warning: unused function 'unicode_cpts_to_utf8' [-Wunused-function]
   29 | static std::string unicode_cpts_to_utf8(const std::vector<uint32_t> & cps) {
      |                    ^~~~~~~~~~~~~~~~~~~~
1 warning generated when compiling for host.
[ 91%] Building CXX object src/CMakeFiles/llama.dir/unicode-data.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/../include -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -MD -MT src/CMakeFiles/llama.dir/unicode-data.cpp.o -MF CMakeFiles/llama.dir/unicode-data.cpp.o.d -o CMakeFiles/llama.dir/unicode-data.cpp.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/unicode-data.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
1 warning generated when compiling for host.
[ 92%] Linking CXX shared library libllama.so
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/src && /usr/bin/cmake -E cmake_link_script CMakeFiles/llama.dir/link.txt --verbose=1
/usr/bin/hipcc -fPIC -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -Wl,-z,relro -Wl,--as-needed  -Wl,-z,pack-relative-relocs -Wl,-z,now -Wl,-z,now   -Wl,--build-id=sha1 -specs=/usr/lib/rpm/redhat/redhat-package-notes -shared -Wl,-soname,libllama.so.b4094 -o libllama.so.b4094 CMakeFiles/llama.dir/llama.cpp.o "CMakeFiles/llama.dir/llama-vocab.cpp.o" "CMakeFiles/llama.dir/llama-grammar.cpp.o" "CMakeFiles/llama.dir/llama-sampling.cpp.o" CMakeFiles/llama.dir/unicode.cpp.o "CMakeFiles/llama.dir/unicode-data.cpp.o"  ../ggml/src/libggml.so.b4094 ../ggml/src/ggml-cpu/libggml-cpu.so ../ggml/src/ggml-hip/libggml-hip.so ../ggml/src/libggml-base.so.b4094
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
clang++: warning: argument unused during compilation: '-Xarch_host -fstack-protector-strong' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-Xarch_host -fcf-protection' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/src && /usr/bin/cmake -E cmake_symlink_library libllama.so.b4094 libllama.so.b4094 libllama.so
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
[ 92%] Built target llama
/usr/bin/gmake  -f common/CMakeFiles/common.dir/build.make common/CMakeFiles/common.dir/depend
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build && /usr/bin/cmake -E cmake_depends "Unix Makefiles" /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094 /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/common /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/common/CMakeFiles/common.dir/DependInfo.cmake "--color="
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
/usr/bin/gmake  -f common/CMakeFiles/common.dir/build.make common/CMakeFiles/common.dir/build
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
[ 93%] Building CXX object common/CMakeFiles/common.dir/arg.cpp.o
[ 94%] Building CXX object common/CMakeFiles/common.dir/common.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/common && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/../include -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -MD -MT common/CMakeFiles/common.dir/arg.cpp.o -MF CMakeFiles/common.dir/arg.cpp.o.d -o CMakeFiles/common.dir/arg.cpp.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/arg.cpp
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/common && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/../include -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -MD -MT common/CMakeFiles/common.dir/common.cpp.o -MF CMakeFiles/common.dir/common.cpp.o.d -o CMakeFiles/common.dir/common.cpp.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/common.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/common.cpp:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/common.h:410:26: warning: unused function 'string_split<std::basic_string<char>>' [-Wunused-function]
  410 | std::vector<std::string> string_split<std::string>(const std::string & input, char separator)
      |                          ^~~~~~~~~~~~~~~~~~~~~~~~~
1 warning generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/common.cpp:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/common.h:410:26: warning: unused function 'string_split<std::basic_string<char>>' [-Wunused-function]
  410 | std::vector<std::string> string_split<std::string>(const std::string & input, char separator)
      |                          ^~~~~~~~~~~~~~~~~~~~~~~~~
1 warning generated when compiling for host.
[ 95%] Building CXX object common/CMakeFiles/common.dir/console.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/common && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/../include -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -MD -MT common/CMakeFiles/common.dir/console.cpp.o -MF CMakeFiles/common.dir/console.cpp.o.d -o CMakeFiles/common.dir/console.cpp.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/console.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 96%] Building CXX object common/CMakeFiles/common.dir/json-schema-to-grammar.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/common && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/../include -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -MD -MT common/CMakeFiles/common.dir/json-schema-to-grammar.cpp.o -MF CMakeFiles/common.dir/json-schema-to-grammar.cpp.o.d -o CMakeFiles/common.dir/json-schema-to-grammar.cpp.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/json-schema-to-grammar.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 97%] Building CXX object common/CMakeFiles/common.dir/log.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/common && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/../include -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -MD -MT common/CMakeFiles/common.dir/log.cpp.o -MF CMakeFiles/common.dir/log.cpp.o.d -o CMakeFiles/common.dir/log.cpp.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/log.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 98%] Building CXX object common/CMakeFiles/common.dir/ngram-cache.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/common && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/../include -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -MD -MT common/CMakeFiles/common.dir/ngram-cache.cpp.o -MF CMakeFiles/common.dir/ngram-cache.cpp.o.d -o CMakeFiles/common.dir/ngram-cache.cpp.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/ngram-cache.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/ngram-cache.cpp:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/common.h:410:26: warning: unused function 'string_split<std::basic_string<char>>' [-Wunused-function]
  410 | std::vector<std::string> string_split<std::string>(const std::string & input, char separator)
      |                          ^~~~~~~~~~~~~~~~~~~~~~~~~
1 warning generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/ngram-cache.cpp:2:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/common.h:410:26: warning: unused function 'string_split<std::basic_string<char>>' [-Wunused-function]
  410 | std::vector<std::string> string_split<std::string>(const std::string & input, char separator)
      |                          ^~~~~~~~~~~~~~~~~~~~~~~~~
1 warning generated when compiling for host.
[ 99%] Building CXX object common/CMakeFiles/common.dir/sampling.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/common && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/. -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/src/../include -I/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -MD -MT common/CMakeFiles/common.dir/sampling.cpp.o -MF CMakeFiles/common.dir/sampling.cpp.o.d -o CMakeFiles/common.dir/sampling.cpp.o -c /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/sampling.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/sampling.cpp:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/sampling.h:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/common.h:410:26: warning: unused function 'string_split<std::basic_string<char>>' [-Wunused-function]
  410 | std::vector<std::string> string_split<std::string>(const std::string & input, char separator)
      |                          ^~~~~~~~~~~~~~~~~~~~~~~~~
1 warning generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/sampling.cpp:1:
In file included from /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/sampling.h:5:
/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/common/common.h:410:26: warning: unused function 'string_split<std::basic_string<char>>' [-Wunused-function]
  410 | std::vector<std::string> string_split<std::string>(const std::string & input, char separator)
      |                          ^~~~~~~~~~~~~~~~~~~~~~~~~
1 warning generated when compiling for host.
[100%] Linking CXX static library libcommon.a
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/common && /usr/bin/cmake -P CMakeFiles/common.dir/cmake_clean_target.cmake
cd /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/common && /usr/bin/cmake -E cmake_link_script CMakeFiles/common.dir/link.txt --verbose=1
/usr/bin/ar qc libcommon.a CMakeFiles/common.dir/arg.cpp.o CMakeFiles/common.dir/common.cpp.o CMakeFiles/common.dir/console.cpp.o "CMakeFiles/common.dir/json-schema-to-grammar.cpp.o" CMakeFiles/common.dir/log.cpp.o "CMakeFiles/common.dir/ngram-cache.cpp.o" CMakeFiles/common.dir/sampling.cpp.o "CMakeFiles/build_info.dir/build-info.cpp.o"
/usr/bin/ranlib libcommon.a
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
[100%] Built target common
gmake[1]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build'
/usr/bin/cmake -E cmake_progress_start /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/redhat-linux-build/CMakeFiles 0

+ module purge
+ local _mlredir=0
+ '[' -n '' ']'
+ case " $@ " in
+ '[' 0 -eq 0 ']'
+ _module_raw purge
++ /usr/bin/tclsh /usr/share/Modules/libexec/modulecmd.tcl bash purge
+ eval 'unset __MODULES_LMCONFLICT;
unset ROCM_BIN;
unset _LMFILES_;
unset LOADEDMODULES;
unset ROCM_GPUS;
unset ROCM_LIB;
test 0;'
++ unset __MODULES_LMCONFLICT
++ unset ROCM_BIN
++ unset _LMFILES_
++ unset LOADEDMODULES
++ unset ROCM_GPUS
++ unset ROCM_LIB
++ test 0
+ _mlstatus=0
+ return 0
+ RPM_EC=0
++ jobs -p
+ exit 0
Executing(%install): /bin/sh -e /var/tmp/rpm-tmp.GeBg09
+ umask 022
+ cd /builddir/build/BUILD/llama-cpp-b4094-build
+ '[' /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT '!=' / ']'
+ rm -rf /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT
++ dirname /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT
+ mkdir -p /builddir/build/BUILD/llama-cpp-b4094-build
+ mkdir /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT
+ CFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer '
+ export CFLAGS
+ CXXFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer'
+ export CXXFLAGS
+ FFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -I/usr/lib64/gfortran/modules '
+ export FFLAGS
+ FCFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -I/usr/lib64/gfortran/modules '
+ export FCFLAGS
+ VALAFLAGS=-g
+ export VALAFLAGS
+ RUSTFLAGS='-Copt-level=3 -Cdebuginfo=2 -Ccodegen-units=1 -Cstrip=none -Cforce-frame-pointers=yes -Clink-arg=-specs=/usr/lib/rpm/redhat/redhat-package-notes --cap-lints=warn'
+ export RUSTFLAGS
+ LDFLAGS='-Wl,-z,relro -Wl,--as-needed  -Wl,-z,pack-relative-relocs -Wl,-z,now -Wl,-z,now   -Wl,--build-id=sha1 -specs=/usr/lib/rpm/redhat/redhat-package-notes '
+ export LDFLAGS
+ LT_SYS_LIBRARY_PATH=/usr/lib64:
+ export LT_SYS_LIBRARY_PATH
+ CC=hipcc
+ export CC
+ CXX=hipcc
+ export CXX
+ cd llama.cpp-b4094
+ DESTDIR=/builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT
+ /usr/bin/cmake --install redhat-linux-build
-- Install configuration: "Release"
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/lib64/libggml-cpu.so
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/lib64/libggml-hip.so
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/lib64/libggml.so.b4094
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/lib64/libggml.so
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-cpu.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-alloc.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-backend.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-blas.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-cann.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-cuda.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-kompute.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-metal.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-rpc.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-sycl.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-vulkan.h
-- Up-to-date: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/lib64/libggml.so.b4094
-- Up-to-date: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/lib64/libggml.so
-- Up-to-date: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml.h
-- Up-to-date: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-cpu.h
-- Up-to-date: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-alloc.h
-- Up-to-date: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-backend.h
-- Up-to-date: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-blas.h
-- Up-to-date: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-cann.h
-- Up-to-date: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-cuda.h
-- Up-to-date: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-kompute.h
-- Up-to-date: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-metal.h
-- Up-to-date: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-rpc.h
-- Up-to-date: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-sycl.h
-- Up-to-date: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/ggml-vulkan.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/lib64/libggml-base.so.b4094
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/lib64/libggml-base.so
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/lib64/libllama.so.b4094
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/lib64/libllama.so
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/include/llama.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/lib64/cmake/llama/llama-config.cmake
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/lib64/cmake/llama/llama-version.cmake
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/bin/convert_hf_to_gguf.py
-- Installing: /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/lib/pkgconfig/llama.pc
+ rm -rf '/builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/lib64/libggml_shared.*'
+ rm /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/bin/convert_hf_to_gguf.py
+ /usr/bin/find-debuginfo -j2 --strict-build-id -m -i --build-id-seed b4094-6.fc42 --unique-debug-suffix -b4094-6.fc42.x86_64 --unique-debug-src-base llama-cpp-b4094-6.fc42.x86_64 --run-dwz --dwz-low-mem-die-limit 10000000 --dwz-max-die-limit 110000000 -S debugsourcefiles.list /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094
find-debuginfo: starting
Extracting debug info from 5 files
DWARF-compressing 5 files
dwz: ./usr/lib64/libggml-base.so.b4094-b4094-6.fc42.x86_64.debug: Unknown debugging section .debug_addr
dwz: ./usr/lib64/libggml-cpu.so-b4094-6.fc42.x86_64.debug: Unknown debugging section .debug_addr
dwz: ./usr/lib64/libggml-hip.so-b4094-6.fc42.x86_64.debug: Unknown debugging section .debug_addr
dwz: ./usr/lib64/libggml.so.b4094-b4094-6.fc42.x86_64.debug: Unknown debugging section .debug_addr
dwz: ./usr/lib64/libllama.so.b4094-b4094-6.fc42.x86_64.debug: Unknown debugging section .debug_addr
dwz: Too few files for multifile optimization
sepdebugcrcfix: Updated 0 CRC32s, 5 CRC32s did match.
Creating .debug symlinks for symlinks to ELF files
Copying sources found by 'debugedit -l' to /usr/src/debug/llama-cpp-b4094-6.fc42.x86_64
find-debuginfo: done
+ /usr/lib/rpm/check-buildroot
+ /usr/lib/rpm/redhat/brp-ldconfig
+ /usr/lib/rpm/brp-compress
+ /usr/lib/rpm/redhat/brp-strip-lto /usr/bin/strip
+ /usr/lib/rpm/brp-strip-static-archive /usr/bin/strip
+ /usr/lib/rpm/check-rpaths
+ /usr/lib/rpm/redhat/brp-mangle-shebangs
+ /usr/lib/rpm/brp-remove-la-files
+ env /usr/lib/rpm/redhat/brp-python-bytecompile '' 1 0 -j2
+ /usr/lib/rpm/redhat/brp-python-hardlink
+ /usr/bin/add-determinism --brp -j2 /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT
Scanned 30 directories and 168 files,
               processed 0 inodes,
               0 modified (0 replaced + 0 rewritten),
               0 unsupported format, 0 errors
Reading /builddir/build/BUILD/llama-cpp-b4094-build/SPECPARTS/rpm-debuginfo.specpart
Processing files: llama-cpp-b4094-6.fc42.x86_64
Executing(%license): /bin/sh -e /var/tmp/rpm-tmp.K2e2cJ
+ umask 022
+ cd /builddir/build/BUILD/llama-cpp-b4094-build
+ cd llama.cpp-b4094
+ LICENSEDIR=/builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/share/licenses/llama-cpp
+ export LC_ALL=C.UTF-8
+ LC_ALL=C.UTF-8
+ export LICENSEDIR
+ /usr/bin/mkdir -p /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/share/licenses/llama-cpp
+ cp -pr /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/LICENSE /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/share/licenses/llama-cpp
+ RPM_EC=0
++ jobs -p
+ exit 0
Provides: libggml-base.so.b4094()(64bit) libggml-hip.so()(64bit) libggml.so.b4094()(64bit) libllama.so.b4094()(64bit) llama-cpp = b4094-6.fc42 llama-cpp(x86-64) = b4094-6.fc42
Requires(rpmlib): rpmlib(CompressedFileNames) <= 3.0.4-1 rpmlib(FileDigests) <= 4.6.0-1 rpmlib(PayloadFilesHavePrefix) <= 4.0-1
Requires: ld-linux-x86-64.so.2()(64bit) ld-linux-x86-64.so.2(GLIBC_2.3)(64bit) libamdhip64.so.6()(64bit) libamdhip64.so.6(hip_4.2)(64bit) libamdhip64.so.6(hip_6.0)(64bit) libc.so.6()(64bit) libc.so.6(GLIBC_2.14)(64bit) libc.so.6(GLIBC_2.17)(64bit) libc.so.6(GLIBC_2.2.5)(64bit) libc.so.6(GLIBC_2.3.4)(64bit) libc.so.6(GLIBC_2.32)(64bit) libc.so.6(GLIBC_2.34)(64bit) libc.so.6(GLIBC_2.38)(64bit) libc.so.6(GLIBC_2.4)(64bit) libc.so.6(GLIBC_ABI_DT_RELR)(64bit) libgcc_s.so.1()(64bit) libgcc_s.so.1(GCC_3.0)(64bit) libggml-base.so.b4094()(64bit) libggml-cpu.so()(64bit) libggml-hip.so()(64bit) libggml.so.b4094()(64bit) libhipblas.so.2()(64bit) libm.so.6()(64bit) libm.so.6(GLIBC_2.2.5)(64bit) libm.so.6(GLIBC_2.27)(64bit) libm.so.6(GLIBC_2.29)(64bit) librocblas.so.4()(64bit) libstdc++.so.6()(64bit) libstdc++.so.6(CXXABI_1.3)(64bit) libstdc++.so.6(CXXABI_1.3.11)(64bit) libstdc++.so.6(CXXABI_1.3.13)(64bit) libstdc++.so.6(CXXABI_1.3.2)(64bit) libstdc++.so.6(CXXABI_1.3.3)(64bit) libstdc++.so.6(CXXABI_1.3.5)(64bit) libstdc++.so.6(GLIBCXX_3.4)(64bit) libstdc++.so.6(GLIBCXX_3.4.11)(64bit) libstdc++.so.6(GLIBCXX_3.4.14)(64bit) libstdc++.so.6(GLIBCXX_3.4.15)(64bit) libstdc++.so.6(GLIBCXX_3.4.17)(64bit) libstdc++.so.6(GLIBCXX_3.4.18)(64bit) libstdc++.so.6(GLIBCXX_3.4.19)(64bit) libstdc++.so.6(GLIBCXX_3.4.20)(64bit) libstdc++.so.6(GLIBCXX_3.4.21)(64bit) libstdc++.so.6(GLIBCXX_3.4.22)(64bit) libstdc++.so.6(GLIBCXX_3.4.25)(64bit) libstdc++.so.6(GLIBCXX_3.4.26)(64bit) libstdc++.so.6(GLIBCXX_3.4.29)(64bit) libstdc++.so.6(GLIBCXX_3.4.30)(64bit) libstdc++.so.6(GLIBCXX_3.4.9)(64bit) rtld(GNU_HASH)
Recommends: numactl
Processing files: llama-cpp-devel-b4094-6.fc42.x86_64
Executing(%doc): /bin/sh -e /var/tmp/rpm-tmp.IB0b71
+ umask 022
+ cd /builddir/build/BUILD/llama-cpp-b4094-build
+ cd llama.cpp-b4094
+ DOCDIR=/builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/share/doc/llama-cpp-devel
+ export LC_ALL=C.UTF-8
+ LC_ALL=C.UTF-8
+ export DOCDIR
+ /usr/bin/mkdir -p /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/share/doc/llama-cpp-devel
+ cp -pr /builddir/build/BUILD/llama-cpp-b4094-build/llama.cpp-b4094/README.md /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT/usr/share/doc/llama-cpp-devel
+ RPM_EC=0
++ jobs -p
+ exit 0
Provides: cmake(llama) libggml-cpu.so()(64bit) llama-cpp-devel = b4094-6.fc42 llama-cpp-devel(x86-64) = b4094-6.fc42
Requires(rpmlib): rpmlib(CompressedFileNames) <= 3.0.4-1 rpmlib(FileDigests) <= 4.6.0-1 rpmlib(PayloadFilesHavePrefix) <= 4.0-1
Requires: cmake-filesystem(x86-64) libc.so.6()(64bit) libc.so.6(GLIBC_2.14)(64bit) libc.so.6(GLIBC_2.2.5)(64bit) libc.so.6(GLIBC_2.29)(64bit) libc.so.6(GLIBC_2.3.2)(64bit) libc.so.6(GLIBC_2.3.4)(64bit) libc.so.6(GLIBC_2.32)(64bit) libc.so.6(GLIBC_2.33)(64bit) libc.so.6(GLIBC_2.34)(64bit) libc.so.6(GLIBC_2.4)(64bit) libc.so.6(GLIBC_2.7)(64bit) libc.so.6(GLIBC_ABI_DT_RELR)(64bit) libgcc_s.so.1()(64bit) libgcc_s.so.1(GCC_3.0)(64bit) libggml-base.so.b4094()(64bit) libggml.so.b4094()(64bit) libllama.so.b4094()(64bit) libm.so.6()(64bit) libm.so.6(GLIBC_2.2.5)(64bit) libm.so.6(GLIBC_2.27)(64bit) libm.so.6(GLIBC_2.29)(64bit) libstdc++.so.6()(64bit) libstdc++.so.6(CXXABI_1.3)(64bit) libstdc++.so.6(GLIBCXX_3.4)(64bit) rtld(GNU_HASH)
Processing files: llama-cpp-debugsource-b4094-6.fc42.x86_64
Provides: llama-cpp-debugsource = b4094-6.fc42 llama-cpp-debugsource(x86-64) = b4094-6.fc42
Requires(rpmlib): rpmlib(CompressedFileNames) <= 3.0.4-1 rpmlib(FileDigests) <= 4.6.0-1 rpmlib(PayloadFilesHavePrefix) <= 4.0-1
Processing files: llama-cpp-debuginfo-b4094-6.fc42.x86_64
Provides: debuginfo(build-id) = 130d475e545b1a3c21c0f506e4f580034b245604 debuginfo(build-id) = b18fd07025039bad19f7ced9d86fe73f08287023 debuginfo(build-id) = e10f7e9265ae4512284dc3149ba19c32b17f0471 debuginfo(build-id) = f4e224c6035f1de8a82d0c36832ce2edd3906c48 libggml-base.so.b4094-b4094-6.fc42.x86_64.debug()(64bit) libggml-hip.so-b4094-6.fc42.x86_64.debug()(64bit) libggml.so.b4094-b4094-6.fc42.x86_64.debug()(64bit) libllama.so.b4094-b4094-6.fc42.x86_64.debug()(64bit) llama-cpp-debuginfo = b4094-6.fc42 llama-cpp-debuginfo(x86-64) = b4094-6.fc42
Requires(rpmlib): rpmlib(CompressedFileNames) <= 3.0.4-1 rpmlib(FileDigests) <= 4.6.0-1 rpmlib(PayloadFilesHavePrefix) <= 4.0-1
Recommends: llama-cpp-debugsource(x86-64) = b4094-6.fc42
Processing files: llama-cpp-devel-debuginfo-b4094-6.fc42.x86_64
Provides: debuginfo(build-id) = 7fe120cf33003f28348e7a69592b038989d5178c libggml-cpu.so-b4094-6.fc42.x86_64.debug()(64bit) llama-cpp-devel-debuginfo = b4094-6.fc42 llama-cpp-devel-debuginfo(x86-64) = b4094-6.fc42
Requires(rpmlib): rpmlib(CompressedFileNames) <= 3.0.4-1 rpmlib(FileDigests) <= 4.6.0-1 rpmlib(PayloadFilesHavePrefix) <= 4.0-1
Recommends: llama-cpp-debugsource(x86-64) = b4094-6.fc42
Checking for unpackaged file(s): /usr/lib/rpm/check-files /builddir/build/BUILD/llama-cpp-b4094-build/BUILDROOT
Wrote: /builddir/build/RPMS/llama-cpp-debuginfo-b4094-6.fc42.x86_64.rpm
Wrote: /builddir/build/RPMS/llama-cpp-debugsource-b4094-6.fc42.x86_64.rpm
Wrote: /builddir/build/RPMS/llama-cpp-devel-debuginfo-b4094-6.fc42.x86_64.rpm
Wrote: /builddir/build/RPMS/llama-cpp-devel-b4094-6.fc42.x86_64.rpm
Wrote: /builddir/build/RPMS/llama-cpp-b4094-6.fc42.x86_64.rpm
Executing(rmbuild): /bin/sh -e /var/tmp/rpm-tmp.PRnh5f
+ umask 022
+ cd /builddir/build/BUILD/llama-cpp-b4094-build
+ test -d /builddir/build/BUILD/llama-cpp-b4094-build
+ /usr/bin/chmod -Rf a+rX,u+w,g-w,o-w /builddir/build/BUILD/llama-cpp-b4094-build
+ rm -rf /builddir/build/BUILD/llama-cpp-b4094-build
+ RPM_EC=0
++ jobs -p
+ exit 0
Finish: rpmbuild llama-cpp-b4094-6.fc42.src.rpm
Finish: build phase for llama-cpp-b4094-6.fc42.src.rpm
INFO: chroot_scan: 1 files copied to /var/lib/copr-rpmbuild/results/chroot_scan
INFO: /var/lib/mock/fedora-rawhide-x86_64-1732737731.306783/root/var/log/dnf5.log
INFO: chroot_scan: creating tarball /var/lib/copr-rpmbuild/results/chroot_scan.tar.gz
/bin/tar: Removing leading `/' from member names
INFO: Done(/var/lib/copr-rpmbuild/results/llama-cpp-b4094-6.fc42.src.rpm) Config(child) 196 minutes 23 seconds
INFO: Results and/or logs in: /var/lib/copr-rpmbuild/results
INFO: Cleaning up build root ('cleanup_on_success=True')
Start: clean chroot
INFO: unmounting tmpfs.
Finish: clean chroot
Finish: run
Running RPMResults tool
Package info:
{
    "packages": [
        {
            "name": "llama-cpp",
            "epoch": null,
            "version": "b4094",
            "release": "6.fc42",
            "arch": "src"
        },
        {
            "name": "llama-cpp",
            "epoch": null,
            "version": "b4094",
            "release": "6.fc42",
            "arch": "x86_64"
        },
        {
            "name": "llama-cpp-devel",
            "epoch": null,
            "version": "b4094",
            "release": "6.fc42",
            "arch": "x86_64"
        },
        {
            "name": "llama-cpp-devel-debuginfo",
            "epoch": null,
            "version": "b4094",
            "release": "6.fc42",
            "arch": "x86_64"
        },
        {
            "name": "llama-cpp-debugsource",
            "epoch": null,
            "version": "b4094",
            "release": "6.fc42",
            "arch": "x86_64"
        },
        {
            "name": "llama-cpp-debuginfo",
            "epoch": null,
            "version": "b4094",
            "release": "6.fc42",
            "arch": "x86_64"
        }
    ]
}
RPMResults finished